advertisement

Calculus Early Transcendentals C. Henry Edwards David E. Penney Seventh Edition Pearson Education Limited Edinburgh Gate Harlow Essex CM20 2JE England and Associated Companies throughout the world Visit us on the World Wide Web at: www.pearsoned.co.uk © Pearson Education Limited 2014 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without either the prior written permission of the publisher or a licence permitting restricted copying in the United Kingdom issued by the Copyright Licensing Agency Ltd, Saffron House, 6–10 Kirby Street, London EC1N 8TS. All trademarks used herein are the property of their respective owners. The use of any trademark in this text does not vest in the author or publisher any trademark ownership rights in such trademarks, nor does the use of such trademarks imply any afﬁliation with or endorsement of this book by such owners. ISBN 10: 1-292-02217-5 ISBN 13: 978-1-292-02217-8 British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Printed in the United States of America P E A R S O N C U S T O M L I B R A R Y Table of Contents Chapter 1. Functions, Graphs, and Models C. Henry Edwards/David E. Penney 1 Chapter 2. Prelude to Calculus C. Henry Edwards/David E. Penney 53 Chapter 3. The Derivative C. Henry Edwards/David E. Penney 105 Chapter 4. Additional Applications of the Derivative C. Henry Edwards/David E. Penney 225 Chapter 5. The Integral C. Henry Edwards/David E. Penney 313 Chapter 6. Applications of the Integral C. Henry Edwards/David E. Penney 413 Chapter 7. Techniques of Integration C. Henry Edwards/David E. Penney 515 Chapter 8. Differential Equations C. Henry Edwards/David E. Penney 575 Chapter 9. Polar Coordinates and Parametric Curves C. Henry Edwards/David E. Penney 659 Chapter 10. Infinite Series C. Henry Edwards/David E. Penney 721 Chapter 11. Vectors, Curves, and Surfaces in Space C. Henry Edwards/David E. Penney 817 Chapter 12. Partial Differentiation C. Henry Edwards/David E. Penney 899 Chapter 13. Multiple Integrals C. Henry Edwards/David E. Penney 997 I II Appendices C. Henry Edwards/David E. Penney 1079 T/F Study Guides—Hints & Answers C. Henry Edwards/David E. Penney 1123 Answers to Odd-Numbered Problems C. Henry Edwards/David E. Penney 1143 References for Further Study C. Henry Edwards/David E. Penney 1233 Index 1235 1 Functions, Graphs, and Models T he seventeenth-century French scholar René Descartes is perhaps better remembered today as a philosopher than as a mathematician. But most of us are familiar with the “Cartesian plane” in which the location of a point P is specified by its coordinates (x, y). As a schoolboy Descartes was often perRené Descartes (1596–1650) mitted to sleep late because of allegedly poor health. He claimed that he always thought most clearly about philosophy, science, and mathematics while he was lying comfortably in bed on cold mornings. After graduating from college, where he studied law (apparently with little enthusiasm), Descartes traveled with various armies for a number of years, but more as a gentleman soldier than as a professional military man. In 1637, after finally settling down (in Holland), Descartes published his famous philosophical treatise Discourse on the Method (of Reasoning Well and Seeking Truth in the Sciences). One of three appendices to this work sets forth his new “analytic” approach to geometry. His principal idea (published almost simultaneously by his countryman Pierre de Fermat) was the correspondence between an equation and its graph, generally a curve in the plane. The equation could be used to study the curve and vice versa. Suppose that we want to solve the equation f (x) = 0. Its solutions are the intersection points of the graph of y = f (x) with the x-axis, so an accurate picture of the curve shows the number and approximate locations of the solutions of the equation. For instance, the graph y = x 3 − 3x 2 + 1 has three x-intercepts, showing that the equation x 3 − 3x 2 + 1 = 0 has three real solutions—one between −1 and 0, one between 0 and 1, and one between 2 and 3. A modern graphing calculator or computer program can approximate these solutions more accurately by magnifying the regions in which they are located. For instance, the magnified center region shows that the corresponding solution is x ≈ 0.65. 4 0.2 2 0.1 y 0 y -2 0 -0.1 -4 -0.2 -4 -2 0 x 2 4 0.4 0.6 x 0.8 The graph y = x 3 − 3x 2 + 1 From Chapter 1 of Calculus, Early Transcendentals, Seventh Edition. C. Henry Edwards, David E. Penney. Copyright © 2008 by Pearson Education, Inc. All rights reserved. 1 2 CHAPTER 1 Functions, Graphs, and Models 1.1 FUNCTIONS AND MATHEMATICAL MODELING Calculus is one of the supreme accomplishments of the human intellect. This mathematical discipline stems largely from the seventeenth-century investigations of Isaac Newton (1642–1727) and Gottfried Wilhelm Leibniz (1646–1716). Yet some of its ideas date back to the time of Archimedes (287–212 B . C .) and originated in cultures as diverse as those of Greece, Egypt, Babylonia, India, China, and Japan. Many of the scientific discoveries that have shaped our civilization during the past three centuries would have been impossible without the use of calculus. The principal objective of calculus is the analysis of problems of change (of motion, for example) and of content (the computation of area and volume, for instance). These problems are fundamental because we live in a world of ceaseless change, filled with bodies in motion and phenomena of ebb and flow. Consequently, calculus remains a vibrant subject, and today this body of conceptual understanding and computational technique continues to serve as the principal quantitative language of science and technology. r Functions Most applications of calculus involve the use of real numbers or variables to describe changing quantities. The key to the mathematical analysis of a geometric or scientific situation is typically the recognition of relationships among the variables that describe the situation. Such a relationship may be a formula that expresses one variable as a function of another. For example: FIGURE 1.1.1 Circle: area A = πr 2, circumference C = 2πr . • The area A of a circle of radius r is given by A = πr 2 (Fig. 1.1.1). The volume V and surface area S of a sphere of radius r are given by V = 43 πr 3 • S = 4πr 2 , respectively (Fig. 1.1.2). After t seconds (s) a body that has been dropped from rest has fallen a distance s = 12 gt 2 r • FIGURE 1.1.2 Sphere: volume V = 43 πr 3 , surface area S = 4πr 2 . and feet (ft) and has speed v = gt feet per second (ft/s), where g ≈ 32 ft/s2 is gravitational acceleration. The volume V (in liters, L) of 3 grams (g) of carbon dioxide at 27◦ C is given in terms of its pressure p in atmospheres (atm) by V = 1.68/ p. DEFINITION Function A real-valued function f defined on a set D of real numbers is a rule that assigns to each number x in D exactly one real number, denoted by f (x). The set D of all numbers for which f (x) is defined is called the domain (or domain of definition) of the function f . The number f (x), read “ f of x,” is called the value of the function f at the number (or point) x. The set of all values y = f (x) is called the range of f . That is, the range of f is the set {y : y = f (x) for some x in D}. In this section we will be concerned more with the domain of a function than with its range. EXAMPLE 1 The squaring function defined by f (x) = x 2 assigns to each real number x its square x 2 . Because every real number can be squared, the domain of f is the set R of all real numbers. But √ only nonnegative numbers are √ squares. Moreover, if a 0, then a = ( a)2 = f ( a), so a is a square. Hence 2 Functions and Mathematical Modeling SECTION 1.1 3 the range of the squaring function f is the set {y : y 0} of all nonnegative real ◗ numbers. Functions can be described in various ways. A symbolic description of the function f is provided by a formula that specifies how to compute the number f (x) in terms of the number x. Thus the symbol f ( ) may be regarded as an operation that is to be performed whenever a number or expression is inserted between the parentheses. EXAMPLE 2 The formula f (x) = x 2 + x − 3 (1) defines a function f whose domain is the entire real line R. Some typical values of f are f (−2) = −1, f (0) = −3, and f (3) = 9. Some other values of the function f are f (4) = 42 + 4 − 3 = 17, f (c) = c2 + c − 3, f (2 + h) = (2 + h)2 + (2 + h) − 3 = (4 + 4h + h 2 ) + (2 + h) − 3 = h 2 + 5h + 3, f (−t 2 ) = (−t 2 )2 + (−t 2 ) − 3 = t 4 − t 2 − 3. x f f (x) FIGURE 1.1.3 A “function machine.” and ◗ When we describe the function f by writing a formula y = f (x), we call x the independent variable and y the dependent variable because the value of y depends— through f —upon the choice of x. As the independent variable x changes, or varies, then so does the dependent variable y. The way that y varies is determined by the rule of the function f . For example, if f is the function of Eq. (1), then y = −1 when x = −2, y = −3 when x = 0, and y = 9 when x = 3. You may find it useful to visualize the dependence of the value y = f (x) on x by thinking of the function f as a kind of machine that accepts as input a number x and then produces as output the number f (x), perhaps displayed or printed (Fig. 1.1.3). One such machine is the square root key of a simple pocket calculator. When a nonnegative number x is entered √ and this key is pressed, the calculator displays (an x. Note that the domain of this square root function approximation to) the number √ f (x) = x is the set [0, +∞) of all nonnegative real numbers, because no negative number has a real square root. √ The range of f is also the set of all nonnegative real numbers, because the symbol x always denotes the nonnegative square root of x. The calculator illustrates its “knowledge” of the domain by displaying an error message if we ask it to calculate the square root of a negative number (or perhaps a complex number, if it’s a more sophisticated calculator). EXAMPLE 3 Not √ every function has a rule expressible as a simple one-part formula such as f (x) = x. For instance, if we write x2 if x 0, h(x) = √ −x if x < 0, then we have defined a perfectly good function with domain R. Some of its values are h(−4) = 2, h(0) = 0, and h(2) = 4. By contrast, the function g in Example 4 is defined initially by means of a verbal description rather than by means of formulas. ◗ EXAMPLE 4 For each real number x, let g(x) denote the greatest integer that is less than or equal to x. For instance, g(2.5) = 2, g(0) = 0, g(−3.5) = −4, and g(π) = 3. If n is an integer, then g(x) = n for every number x such that n x < n + 1. This function g is called the greatest integer function and is often denoted by g(x) = [[x]]. 3 4 CHAPTER 1 Functions, Graphs, and Models Thus [[2.5]] = 2, [[−3.5]] = −4, and [[π ]] = 3. Note that although [[x]] is defined for all x, the range of the greatest integer function is not all of R, but the set Z of all ◗ integers. The name of a function need not be a single letter such as f or g. For instance, think of the trigonometric functions sin(x) and cos(x) with the names sin and cos. EXAMPLE 5 Another descriptive name for the greatest integer function of Example 4 is F LOOR(x) = [[x]]. (2) (We think of the integer n as the “floor” beneath the real numbers lying between n and n + 1.) Similarly, we may use ROUND(x) to name the familiar function that “rounds off” the real number x to the nearest integer n, except that ROUND(x) = n + 1 if x = n + 12 (so we “round upward” in case of ambiguity). Round off enough different numbers to convince yourself that (3) ROUND(x) = F LOOR x + 12 for all x. Closely related to the F LOOR and ROUND functions is the “ceiling function” used by the U.S. Postal Service; C EILING(x) denotes the least integer that is not less than the number x. In 2006 the postage rate for a first-class letter was 39/ c for the first ounce and 24/ c for each additional ounce or fraction thereof. For a letter weighing w > 0 ounces, the number of “additional ounces” involved is C EILING(w) − 1. Therefore the postage s(w) due on this letter is given by s(w) = 39 + 24 · [C EILING(w) − 1] = 15 + 24 · C EILING(w). ◗ Domains and Intervals The function f and the value or expression f (x) are different in the same sense that a machine and its output are different. Nevertheless, it is common to use an expression like “the function f (x) = x 2 ” to define a function merely by writing its formula. In this situation the domain of the function is not specified. Then, by convention, the domain of the function f is the set of all real numbers x for which the expression f (x) makes sense and produces a real number y. For instance, the domain of the function h(x) = 1/x is the set of all nonzero real numbers (because 1/x is defined precisely when x = 0). (1, 3) An open interval A closed interval [−1, 2] A half-open interval [0, 1.5) A half-open interval (−1, 1] [ 12 , ∞) An unbounded interval An unbounded interval (−∞, 2) FIGURE 1.1.4 Some examples of intervals of real numbers. Domains of functions frequently are described in terms of intervals of real numbers (Fig. 1.1.4). (Interval notation is reviewed in Appendix A.) Recall that a closed interval [a, b] contains both its endpoints x = a and x = b, whereas the open interval (a, b) contains neither endpoint. Each of the half-open intervals [a, b) and (a, b] contains exactly one of its two endpoints. The unbounded interval [a, ∞) contains its endpoint x = a, whereas (−∞, a) does not. The previously mentioned domain of h(x) = 1/x is the union of the unbounded intervals (−∞, 0) and (0, ∞). 4 Functions and Mathematical Modeling SECTION 1.1 5 (−∞, −2) EXAMPLE 6 Find the domain of the function g(x) = (−2, ∞) −2 0 FIGURE 1.1.5 The domain of g(x) = 1/(2x + 4) is the union of two unbounded open intervals. 1 . 2x + 4 Solution Division by zero is not allowed, so the value g(x) is defined precisely when 2x + 4 = 0. This is true when 2x = −4, and thus when x = −2. Hence the domain of g is the set {x : x = 2}, which is the union of the two unbounded open intervals ◗ (−∞, −2) and (−2, ∞), shown in Fig. 1.1.5. 1 EXAMPLE 7 Find the domain of h(x) = √ . 2x + 4 Solution Now it is necessary not only that√ the quantity 2x +4 be nonzero, but also that it be positive, in order that the square root 2x + 4 is defined. But 2x + 4 > 0 when 2x > −4, and thus when x > −2. Hence the domain of h is the single unbounded ◗ open interval (−2, ∞). Mathematical Modeling The investigation of an applied problem often hinges on defining a function that captures the essence of a geometrical or physical situation. Examples 8 and 9 illustrate this process. EXAMPLE 8 A rectangular box with a square base has volume 125. Express its total surface area A as a function of the edge length x of its base. y x x FIGURE 1.1.6 The box of Example 8. Solution The first step is to draw a sketch and to label the relevant dimensions. Figure 1.1.6 shows a rectangular box with square base of edge length x and with height y. We are given that the volume of the box is V = x 2 y = 125. (4) Both the top and the bottom of the box have area x 2 and each of its four vertical sides has area x y, so its total surface area is A = 2x 2 + 4x y. (5) But this is a formula for A in terms of the two variables x and y rather than a function of the single variable x. To eliminate y and thereby obtain A in terms of x alone, we solve Eq. (4) for y = 125/x 2 and then substitute this result in Eq. (5) to obtain 125 500 = 2x 2 + . 2 x x Thus the surface area is given as a function of the edge length x by A = 2x 2 + 4x · 500 , 0 < x < +∞. (6) x It is necessary to specify the domain because negative values of x make sense in the formula in (5) but do not belong in the domain of the function A. Because every x > 0 determines such a box, the domain does, in fact, include all positive real numbers. ◗ A(x) = 2x 2 + COMMENT In Example 8 our goal was to express the dependent variable A as a function of the independent variable x. Initially, the geometric situation provided us instead with 1. The formula in Eq. (5) expressing A in terms of both x and the additional variable y, and 2. The relation in Eq. (4) between x and y, which we used to eliminate y and thereby express A as a function of x alone. We will see that this is a common pattern in many different applied problems, such as the one that follows. 5 6 CHAPTER 1 Functions, Graphs, and Models x $5/ft y $5/ft $5/ft y $1/ft x FIGURE 1.1.7 The animal pen. Wall The Animal Pen Problem You must build a rectangular holding pen for animals. To save material, you will use an existing wall as one of its four sides. The fence for the other three sides costs $5/ft, and you must spend $1/ft to paint the portion of the wall that forms the fourth side of the pen. If you have a total of $180 to spend, what dimensions will maximize the area of the pen you can build? Figure 1.1.7 shows the animal pen and its dimensions x and y, along with the cost per foot of each of its four sides. When we are confronted with a verbally stated applied problem such as this, our first question is, How on earth do we get started on it? The function concept is the key to getting a handle on such a situation. If we can express the quantity to be maximized—the dependent variable—as a function of some independent variable, then we have something tangible to do: Find the maximum value attained by the function. Geometrically, what is the highest point on that function’s graph? EXAMPLE 9 In connection with the animal pen problem, express the area A of the pen as a function of the length x of its wall side. Solution The area A of the rectangular pen of length x and width y is A = x y. (7) When we multiply the length of each side in Fig. 1.1.7 by its cost per foot and then add the results, we find that the total cost C of the pen is C = x + 5y + 5x + 5y = 6x + 10y. So 6x + 10y = 180, (8) because we are given C = 180. Choosing x to be the independent variable, we use the relation in Eq. (8) to eliminate the additional variable y from the area formula in Eq. (7). We solve Eq. (8) for y and substitute the result y= 1 (180 10 − 6x) = 35 (30 − x) (9) in Eq. (7). Thus we obtain the desired function A(x) = 35 (30x − x 2 ) that expresses the area A as a function of the length x. In addition to this formula for the function A, we must also specify its domain. Only if x > 0 will actual rectangles be produced, but we find it convenient to include the value x = 0 as well. This value of x corresponds to a “degenerate rectangle” of base length zero and height y= 3 5 · 30 = 18, a consequence of Eq. (9). For similar reasons, we have the restriction y 0. Because y = 35 (30 − x), it follows that x 30. Thus the complete definition of the area function is A(x) = 35 (30x − x 2 ), 0 x 30. (10) ◗ COMMENT The domain of a function is a necessary part of its definition, and for each function we must specify the domain of values of the independent variable. In applications, we use the values of the independent variable that are relevant to the problem at hand. 6 Functions and Mathematical Modeling SECTION 1.1 7 x Example 9 illustrates an important part of the solution of a typical applied problem—the formulation of a mathematical model of the physical situation under study. The area function A(x) defined in (10) provides a mathematical model of the animal pen problem. The shape of the optimal animal pen can be determined by finding the maximum value attained by the function A on its domain of definition. A(x) 0 5 10 15 20 25 30 0 75 120 135← 120 75 0 Numerical Investigation Armed with the result of Example 9, we might attack the animal pen problem by calculating a table of values of the area function A(x) in Eq. (10). Such a table is shown in Fig. 1.1.8. The data in this table suggest strongly that the maximum area is A = 135 ft2 , attained with side length x = 15 ft, in which case Eq. (9) yields y = 9 ft. This conjecture appears to be corroborated by the more refined data shown in Fig. 1.1.9. Thus it seems that the animal pen with maximal area (costing $180) is x = 15 ft long and y = 9 ft wide. The tables in Figs. 1.1.8 and 1.1.9 show only integral values of x, however, and it is quite possible that the length x of the pen of maximal area is not an integer. Consequently, numerical tables alone do not settle the matter. A new mathematical idea is needed in order to prove that A(15) = 135 is the maximum value of A(x) = 35 (30x − x 2 ), 0 x 30 FIGURE 1.1.8 Area A(x) of a pen with side of length x. x A(x) 10 11 12 13 14 15 16 17 18 19 20 120 125.4 129.6 132.6 134.4 135 ← 134.4 132.6 129.6 125.4 120 for all x in its domain. We attack this problem again in Section 1.2. Tabulation of Functions Many scientific and graphing calculators allow the user to program a given function for repeated evaluation, and thereby to painlessly compute tables like those in Figs. 1.1.8 and 1.1.9. For instance, Figs. 1.1.10 and 1.1.11 show displays of a calculator prepared to calculate values of the dependent variable y1 = A(x) = (3/5)(30x − x 2 ), and Fig. 1.1.12 shows the calculator’s resulting version of the table in Fig. 1.1.9. The use of a calculator or computer to tabulate values of a function is a simple technique with surprisingly many applications. Here we illustrate a method of solving approximately an equation of the form f (x) = 0 by repeated tabulation of values f (x) of the function f . As a specific example, suppose that we ask what value of x in Eq. (10) yields an animal pen of area A = 100. Then we need to solve the equation FIGURE 1.1.9 Further indication that x = 15 yields maximal area A = 135. A(x) = 35 (30x − x 2 ) = 100, which is equivalent to the equation f (x) = 35 (30x − x 2 ) − 100 = 0. (11) This is a quadratic equation that could be solved using the quadratic formula of basic algebra, but we want to take a more direct, numerical approach. The reason is that the t t TEXAS INSTRUMENTS TI-83 FIGURE 1.1.10 A calculator programmed to evaluate A(x) = (3/5)(30x − x 2 ). t TEXAS INSTRUMENTS TEXAS INSTRUMENTS TI-83 TI-83 FIGURE 1.1.11 The table setup. FIGURE 1.1.12 The resulting table. 7 8 CHAPTER 1 Functions, Graphs, and Models numerical approach is applicable even when no simple formula (such as the quadratic formula) is available. The data in Fig. 1.1.8 suggest that one value of x for which A(x) = 100 lies somewhere between x = 5 and x = 10 and that a second such value lies between x = 20 and x = 25. Indeed, substitution in Eq. (11) yields f (5) = −25 < 0 and f (10) = 20 > 0. The fact that f (x) is negative at one endpoint of the interval [5, 10] but positive at the other endpoint suggests that f (x) is zero somewhere between x = 5 and x = 10. To see where, we tabulate values of f (x) on [5, 10]. In the table of Fig. 1.1.13 we see that f (7) < 0 and f (8) > 0, so we focus next on the interval [7, 8]. Tabulating f (x) on [7, 8] gives the table of Fig. 1.1.14, where we see that f (7.3) < 0 and f (7.4) > 0. We therefore tabulate f (x) once more, this time on the interval [7.3, 7.4]. In Fig. 1.1.15 we see that f (7.36) ≈ −0.02 and f (7.37) ≈ 0.07. Because f (7.36) is considerably closer to zero than is f (7.37), we conclude that the desired solution of Eq. (11) is given approximately by x ≈ 7.36, accurate to two decimal places. If greater accuracy were needed, we could continue to tabulate f (x) on smaller and smaller intervals. If we were to begin with the interval [20, 25] and proceed similarly, we would find the second value x ≈ 22.64 such that f (x) = 0. (You should do this for practice.) Finally, let’s calculate the corresponding values of the width y of the animal pen such that A = x y = 100: • • If x ≈ 7.36, then y ≈ 13.59. If x ≈ 22.64, then y ≈ 4.42. Thus, under the cost constraint of the animal pen problem, we can construct either a 7.36-ft by 13.59-ft or a 22.64-ft by 4.42-ft rectangle, both of area 100 ft2 . The layout of Figs. 1.1.13 through 1.1.15 suggests the idea of repeated tabulation as a process of successive numerical magnification. This method of repeated tabulation can be applied to a wide range of equations of the form f (x) = 0. If the interval [a, b] contains a solution and the endpoint values f (a) and f (b) differ in sign, then we can approximate this solution by tabulating values on successively smaller subintervals. Problems 57 through 66 and the project at the end of this section are applications of this concrete numerical method for the approximate solution of equations. x f (x) 5 6 7 8 9 10 ⫺25.0 ⫺13.6 ⫺3.4 5.6 13.4 20.0 FIGURE 1.1.13 Values of f (x) on [5, 10]. x f (x) x 7.0 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 8.0 ⫺3.400 ⫺2.446 ⫺1.504 ⫺0.574 0.344 1.250 2.144 3.026 3.896 4.754 5.600 7.30 7.31 7.32 7.33 7.34 7.35 7.36 7.37 7.38 7.39 7.40 FIGURE 1.1.14 Values of f (x) on [7, 8]. 8 f(x) ⫺0.5740 ⫺0.4817 ⫺0.3894 ⫺0.2973 ⫺0.2054 ⫺0.1135 ⫺0.0218 0.0699 0.1614 0.2527 0.3440 FIGURE 1.1.15 Values of f (x) on [7.3, 7.4]. Functions and Mathematical Modeling SECTION 1.1 9 1.1 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Isaac Newton was born in the 18th century. 2. A function is a rule that assigns to each real number in its domain one and only one real number. 3. The value of the function f at the number x in its domain is commonly denoted by f (x). 4. If the domain of the function f is not specified, then it is the set of all real numbers. 5. The function giving the surface area A as a function of the edge length x of the box of Example 8 is given by A(x) = 2x 2 + 600 , x 0 x < +∞. 6. In the animal pen problem (Example 9), the maximum area is attained when the length x of the wall side is 18 ft. 7. The interval (a, b) is said to be open because it contains neither of its endpoints a and b. √ 8. The domain of f (x) = x does not include the number x = −4. 9. The domain of the function A(x) = 35 (30x − x 2 ) is the set of all real numbers. 10. There is no good reason why the domain of the animal pen function in Eq. (10) is restricted to the interval 0 x 30. 1.1 CONCEPTS: QUESTIONS AND DISCUSSION 1. Can a function have the same value at two different points? Can it have two different values at the same point x? 2. Explain the difference between a dependent variable and an independent variable. A change in one both causes and determines a change in the other. Which one is the “controlling variable”? 3. What is the difference between an open interval and a closed interval? Is every interval on the real line either open or closed? Justify your answer. 4. Suppose that S is a set of real numbers. Is there a function whose domain of definition is precisely the set S? Is there a function defined on the whole real line whose range is precisely the set S? Is there a function that has the value 1 at each point of S and the value 0 at each point of the real line R not in S? 5. Figure 1.1.6 shows a box with square base and height y. Which of the following two formulas would suffice to define the volume V of this box as a function of y? (a) V = x 2 y; (b) V = y(10 − 2y)2 . Discuss the difference between a formula and a function. 6. In the following table, y is a function of x. Determine whether or not x is a function of y. x 0 2 4 6 8 10 y −1 3 8 7 3 −2 9 10 CHAPTER 1 Functions, Graphs, and Models 1.1 PROBLEMS In Problems 1 through 4, find and simplify √ each of the following values: (a) f (−a); (b) f (a −1 ); (c) f ( a); (d) f (a 2 ). 1. f (x) = 3. f (x) = 1 x 2. f (x) = x 2 + 5 x2 1 +5 4. f (x) = √ 1 + x2 + x4 In Problems 5 through 10, find all values of a such that g(a) = 5. 5. g(x) = 3x + 4 √ 7. g(x) = x 2 + 16 √ 9. g(x) = 3 x + 25 6. g(x) = 1 2x − 1 38. Express the volume V of a sphere as a function of its surface area S. 39. Given: 0◦ C is the same as 32◦ F, and a temperature change of 1◦ C is the same as a change of 1.8◦ F. Express the Celsius temperature C as a function of the Fahrenheit temperature F. 40. Show that if a rectangle has base x and perimeter 100 (Fig. 1.1.16), then its area A is given by the function A(x) = x(50 − x), 0 x 50. 8. g(x) = x 3 − 3 10. g(x) = 2x 2 − x + 4 y In Problems 11 through 16, compute and then simplify the quantity f (a + h) − f (a). 11. f (x) = 3x − 2 12. f (x) = 1 − 2x 13. f (x) = x 1 15. f (x) = x 14. f (x) = x 2 + 2x 2 16. f (x) = x +1 2 In Problems 17 through 20, find the range of values of the given function. ⎧ x ⎨ if x = 0; 17. f (x) = |x| ⎩ 0 if x = 0 18. f (x) = [[3x]] exceeding x.) x FIGURE 1.1.16 A = x y (Problem 40). 41. A rectangle with base of length x is inscribed in a circle of radius 2 (Fig 1.1.17). Express the area A of the rectangle as a function of x. (Recall that [[x]] is the largest integer not y 2 19. f (x) = (−1)[[x]] 20. f (x) is the first-class postage (in cents) for a letter mailed in the United States and weighing x ounces, 0 < x < 12. As of January 8, 2006 the postage rate for such a letter was 39/ c for the first ounce plus 24/ c for each additional ounce or fraction thereof. In Problems 21 through 35, find the largest domain (of real numbers) on which the given formula determines a (real-valued) function. 21. f (x) = 10 − x 2 √ 23. f (t) = t 2 √ 25. f (x) = 3x − 5 √ 27. f (t) = 1 − 2t 2 3−x √ 31. f (x) = x 2 + 9 29. f (x) = 33. f (x) = 4− √ x 22. f (x) = x 3 + 5 √ 2 24. g(t) = t √ 3 26. g(t) = t + 4 1 28. g(x) = (x + 2)2 2 30. g(t) = 3−t 1 32. h(z) = √ 4 − z2 x +1 34. f (x) = x −1 t |t| 36. Express the area A of a square as a function of its perimeter P. 2 x 2 FIGURE 1.1.17 A = x y (Problem 41). 42. An oil field containing 20 wells has been producing 4000 barrels of oil daily. For each new well that is drilled, the daily production of each well decreases by 5 barrels per day. Write the total daily production of the oil field as a function of the number x of new wells drilled. 43. Suppose that a rectangular box has volume 324 cm3 and a square base of edge length x centimeters. The material for the base of the box costs 2/ c/cm2 , and the material for its top and four sides costs 1/ c/cm2 . Express the total cost of the box as a function of x. See Fig. 1.1.18. y 35. g(t) = 37. Express the circumference C of a circle as a function of its area A. 10 x x FIGURE 1.1.18 V = x 2 y (Problem 43). Functions and Mathematical Modeling SECTION 1.1 11 44. A rectangle of fixed perimeter 36 is rotated around one of its sides S to generate a right circular cylinder. Express the volume V of this cylinder as a function of the length x of the side S. See Fig. 1.1.19. x FIGURE 1.1.22 The box of Problem 47. y x FIGURE 1.1.19 V = π x y 2 (Problem 44). 45. A right circular cylinder has volume 1000 in.3 and the radius of its base is r inches. Express the total surface area A of the cylinder as a function of r . See Fig. 1.1.20. 48. Continue Problem 40 by numerically investigating the area of a rectangle of perimeter 100. What dimensions (length and width) would appear to maximize the area of such a rectangle? 49. Determine numerically the number of new oil wells that should be drilled to maximize the total daily production of the oil field of Problem 42. 50. Investigate numerically the total surface area A of the rectangular box of Example 8. Assuming that both x 1 and y 1, what dimensions x and y would appear to minimize A? Problems 51 through 56 deal with the functions C EILING, F LOOR, and ROUND of Example 5. 51. Show that C EILING(x) = −F LOOR(−x) for all x. 52. Suppose that k is a constant. What is the range of the function g(x) = ROUND(kx)? 53. What is the range of the function g(x) = 101 ROUND(10x)? h r FIGURE 1.1.20 V = πr 2 h (Problem 45). 46. A rectangular box has total surface area 600 cm2 and a square base with edge length x centimeters. Express the volume V of the box as a function of x. 47. An open-topped box is to be made from a square piece of cardboard of edge length 50 in. First, four small squares, each of edge length x inches, are cut from the corners of the cardboard (Fig. 1.1.21). Then the four resulting flaps are turned up—folded along the dotted lines—to form the four sides of the box, which will thus have a square base and a depth of x inches (Fig. 1.1.22). Express its volume V as a function of x. 50 50 x x ? x FIGURE 1.1.21 Fold the edges up to make a box (Problem 47). 1 ROUND(100π) = 54. Recalling that π ≈ 3.14159, note that 100 3.14. Hence define (in terms of ROUND) a function ROUND 2(x) that gives the value of x rounded accurate to two decimal places. 55. Define a function ROUND 4(x) that gives the value of x rounded accurate to four decimal places, so that ROUND 4(π) = 3.1416. 56. Define a function C HOP 4(x) that “chops off” (or discards) all decimal places of x beyond the fourth one, so that C HOP 4(π ) = 3.1415. In Problems 57 through 66, a quadratic equation ax 2 +bx+c = 0 and an interval [ p, q] containing one of its solutions are given. Use the method of repeated tabulation to approximate this solution with two digits correct or correctly rounded to the right of the decimal. Check that your result agrees with one of the two solutions given by the quadratic formula, √ −b ± b2 − 4ac x= . 2a 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. x 2 − 3x + 1 = 0, [0, 1] x 2 − 3x + 1 = 0, [2, 3] x 2 + 2x − 4 = 0, [1, 2] x 2 + 2x − 4 = 0, [−4, −3] 2x 2 − 7x + 4 = 0, [0, 1] 2x 2 − 7x + 4 = 0, [2, 3] x 2 − 11x + 25 = 0, [3, 4] x 2 − 11x + 25 = 0, [7, 8] 3x 2 + 23x − 45 = 0, [1, 2] 3x 2 + 23x − 45 = 0, [−10, −9] 11 12 CHAPTER 1 Functions, Graphs, and Models 1.1 INVESTIGATION: Designing a Wading Pool Starting with a given rectangular piece of tin, you are to design a wading pool in the manner indicated by Figs. 1.1.21 and 1.1.22. Your task is to investigate the maximal volume pool that can be constructed, and how to construct a wading pool of specified volume. For your own personal wading pool, start with a square piece of tin of size a × b feet, where a and b < a are the two largest digits in your student ID number. You need to determine the corner notch edge length x so that the wading pool you construct will have the largest possible volume V . Start by expressing the box’s volume V = f (x) as a function of its height x, and then use the method of repeated tabulation to find the maximum value Vmax (rounded off accurate to 2 decimal places) attained by the function f (x) on the interval [0, b/2]. (Why is this the appropriate domain of f ?) For a second investigation, suppose you decide instead that you want your pool to have exactly half the maximum possible volume Vmax . Note first that a tabulation of f (x) on the interval [0, b/2] indicates that this is true for two different values of x. Find each of them (rounded off accurate to 3 decimal places). Write up the results of your investigations in the form of a carefully organized report consisting of complete sentences (plus pertinent equations and data tables) explaining your results in detail, and telling precisely what you did to solve your problems. 1.2 GRAPHS OF EQUATIONS AND FUNCTIONS Graphs and equations of straight lines in the x y-coordinate plane are reviewed in Appendix B. Recall the slope-intercept equation y y = mx + b y = mx + b (1) of the straight line with slope m = tan φ, angle of inclination φ, and y-intercept b (Fig. 1.2.1). The “rise over run” definition φ b m= x FIGURE 1.2.1 A line with y-intercept b and inclination angle φ. y y2 − y1 rise = = run x x2 − x1 (2) of the slope (Fig. 1.2.2) leads to the point-slope equation y − y0 = m(x − x0 ) (3) of the straight line with slope m that passes through the point (x0 , y0 )—see Fig. 1.2.3. In either case a point (x, y) in the x y-plane lies on the line if and only if its coordinates x and y satisfy the indicated equation. y (x2, y2 ) y Δy = y2 − y1 (x1, y1) (x0, y0 ) φ y − y0 = m(x − x0 ) Δx = x2 − x1 (x, y) x x FIGURE 1.2.2 Slope y m = tan φ = . x 12 FIGURE 1.2.3 The line through (x0 , y0 ) with slope m. Graphs of Equations and Functions SECTION 1.2 13 If y = 0 in Eq. (2), then m = 0 and the line is horizontal. If x = 0, then the line is vertical and (because we cannot divide by zero) the slope of the line is not defined. Thus: y φ • • Horizontal lines have slope zero. Vertical lines have no defined slope at all. φ x EXAMPLE 1 Write an equation of the line L that passes through the point P(3, 5) and is parallel to the line having equation y = 2x − 4. Solution The two parallel lines have the same angle of inclination φ (Fig. 1.2.4) and therefore have the same slope m. Comparing the given equation y = 2x − 4 with the slope-intercept equation in (1), we see that m = 2. The point-slope equation therefore gives y − 5 = 2(x − 3) FIGURE 1.2.4 Parallel lines have the same slope m = tan φ. y —alternatively, y = 2x − 1, for an equation of the line L. ◗ Equations (1) and (3) can both be put into the form of the general linear equation x FIGURE 1.2.5 The graph of the equation x 2 + y 2 = (x 2 + y 2 − 2x)2 . y A x + By = C. (4) Conversely, if B = 0, then we can divide the terms in Eq. (4) by B and solve for y, thereby obtaining the slope-intercept equation of a straight line. If A = 0, then the resulting equation has the form y = H , the equation of a horizontal line with slope zero. If B = 0 but A = 0, then Eq. (4) can be solved for x = K , the equation of a vertical line (having no slope at all). In summary, we see that if the coefficients A and B are not both zero, then Eq. (4) is the equation of some straight line in the plane. P2 (x2, y2 ) Graphs of More General Equations d P1(x1, y1) A straight line is a simple example of the graph of an equation. By contrast, a computer-graphing program produced the exotic curve shown in Fig. 1.2.5 when asked to picture the set of all points (x, y) satisfying the equation y 2 − y1 x2 − x1 x x 2 + y 2 = (x 2 + y 2 − 2x)2 . Both a straight line and this complicated curve are examples of graphs of equations. FIGURE 1.2.6 The Pythagorean theorem implies the distance formula d = (x2 − x1 )2 + (y2 − y1 )2 . DEFINITION Graph of an Equation The graph of an equation in two variables x and y is the set of all points (x, y) in the plane that satisfy the equation. For example, the distance formula of Fig. 1.2.6 tells us that the graph of the equation y x 2 + y2 = r 2 (5) (x, y) r is the circle of radius r centered at the origin (0, 0). More generally, the graph of the equation (h, k) (x − h)2 + (y − k)2 = r 2 (6) x FIGURE 1.2.7 A translated circle. is the circle of radius r with center (h, k). This also follows from the distance formula, because the distance between the points (x, y) and (h, k) in Fig. 1.2.7 is r . 13 14 CHAPTER 1 Functions, Graphs, and Models EXAMPLE 2 The equation of the circle with center (3, 4) and radius 10 is (x − 3)2 + (y − 4)2 = 100, which may also be written in the form x 2 + y 2 − 6x − 8y − 75 = 0. ◗ Translates of Graphs Suppose that the x y-plane is shifted rigidly (or translated) by moving each point h units to the right and k units upward. (A negative value of h or k corresponds to a leftward or downward movement.) That is, each point (x, y) of the plane is moved to the point (x + h, y + k); see Fig. 1.2.8. Then the circle with radius r and center (0, 0) is translated to the circle with radius r and center (h, k). Thus the general circle described by Eq. (6) is a translate of the origin-centered circle. Note that the equation of the translated circle is obtained from the original equation by replacing x with x − h and y with y − k. This observation illustrates a general principle that describes equations of translated (or “shifted”) graphs. y (x + h, y + k) (x, y) x FIGURE 1.2.8 Translating a point. Translation Principle When the graph of an equation is translated h units to the right and k units upward, the equation of the translated curve is obtained from the original equation by replacing x with x − h and y with y − k. Observe that we can write the equation of a translated circle in Eq. (6) in the general form x 2 + y 2 + ax + by = c. (7) What, then, can we do when we encounter an equation already of the form in Eq. (7)? We first recognize that the graph is likely to be a circle. If so, we can discover its center and radius by the technique of completing the square. To do so, we note that x 2 + ax = x + a 2 2 − a2 , 4 which shows that x 2 + ax can be made into the perfect square (x + 12 a)2 by adding to it the square of half the coefficient of x. EXAMPLE 3 Find the center and radius of the circle that has the equation x 2 + y 2 − 4x + 6y = 12. 8 Solution We complete the square separately for each of the variables x and y. This 4 gives y 0 (x 2 − 4x + 4) + (y 2 + 6y + 9) = 12 + 4 + 9; (2, −3) -4 (x − 2)2 + (y + 3)2 = 25. -8 -10 -5 0 x 5 FIGURE 1.2.9 The circle of Example 3. 10 Hence the circle—shown in Fig. 1.2.9—has center (2, −3) and radius 5. Solving the last equation for y gives y = −3 ± 25 − (x − 2)2 . Thus the whole circle consists of the graphs of the two equations and y = −3 + 25 − (x − 2)2 y = −3 − 25 − (x − 2)2 that describe its upper and lower semicircles. 14 ◗ Graphs of Equations and Functions SECTION 1.2 15 Graphs of Functions The graph of a function is a special case of the graph of an equation. DEFINITION Graph of a Function The graph of the function f is the graph of the equation y = f (x). Thus the graph of the function f is the set of all points in the plane that have the form (x, f (x)), where x is in the domain of f . (See Fig. 1.2.10.) Because the second coordinate of such a point is uniquely determined by its first coordinate, we obtain the following useful principle: y (x3, f(x3)) (x1, f (x1)) y = f (x) (x2, f(x2 )) f(x3) f(x2 ) f(x1) x1 x2 x3 x FIGURE 1.2.10 The graph of the function f . The Vertical Line Test Each vertical line through a point in the domain of a function meets its graph in exactly one point. Thus no vertical line can intersect the graph of a function in more than one point. For instance, it follows that the curve in Fig. 1.2.5 cannot be the graph of a function, although it is the graph of an equation. Similarly, a circle cannot be the graph of a function. y 0 y= x< x r fo fo r −x x> 0 y= EXAMPLE 4 Construct the graph of the absolute value function f (x) = |x|. y = |x| Solution Recall that x FIGURE 1.2.11 The graph of the absolute value function y = |x| of Example 4. |x| = x −x if x 0, if x < 0. So the graph of y = |x| consists of the right half of the line y = x together with the ◗ left half of the line y = −x, as shown in Fig. 1.2.11. EXAMPLE 5 Sketch the graph of the reciprocal function f (x) = 1 . x Solution Let’s examine four natural cases. 1. When x is positive and numerically large, f (x) is small and positive. 2. When x is positive and near zero, f (x) is large and positive. 3. When x is negative and numerically small (negative and close to zero), f (x) is large and negative. 4. When x is large and negative (x is negative but |x| is large), f (x) is small and negative (negative and close to zero). 15 16 CHAPTER 1 Functions, Graphs, and Models To get started with the graph, we can plot a few points, such as (1, 1), (−1, −1), 5, 15 , 15 , 5 , −5, − 15 , and − 15 , −5 . y 6 ( 15 , 5) 4 y= −6 −4 (−5, − 15 ) 1 x (1, 1) 2 (5, −2 2 (−1, −1) 1 ) 5 6x 4 −2 −4 (− 15 , −5) −6 The locations of these points, together with the four cases just discussed, suggest that ◗ the actual graph resembles the one shown in Fig. 1.2.12. Figure 1.2.12 exhibits a “gap,” or “discontinuity,” in the graph of y = 1/x at x = 0. Indeed, the gap is called an infinite discontinuity because y increases without bound as x approaches zero from the right, whereas y decreases without bound as x approaches zero from the left. This phenomenon generally is signaled by the presence of denominators that are zero at certain values of x, as in the case of the functions FIGURE 1.2.12 The graph of the reciprocal function y = 1/x of Example 5. f (x) = 1 1−x f (x) = and 1 , x2 which we ask you to graph in the problems. y … 3 2 EXAMPLE 6 Figure 1.2.13 shows the graph of the greatest integer function f (x) = [[x]] in Example 4 in Section 1.1. Note the “jumps” that occur at integral values of x. On calculators, the greatest integer function is sometimes denoted by INT ; in some ◗ programming languages, it is called FLOOR. 1 −3 −2 −1 1 2 3 x EXAMPLE 7 Graph the function with the formula f (x) = x − [[x]] − 12 . −1 −2 … −3 FIGURE 1.2.13 The graph of the greatest integer function f (x) = [[x]] of Example 6. Solution Recall that [[x]] = n, where n is the greatest integer not exceeding x—thus n x < n + 1. Hence if n is an integer, then f (n) = n − n − 1 2 = − 12 . This implies that the point (n, − 12 ) lies on the graph of f for each integer n. Next, if n x < n + 1 (where, again, n is an integer), then f (x) = x − n − 12 . Because y = x −n− 12 has as its graph a straight line of slope 1, it follows that the graph of f takes the form shown in Fig. 1.2.14. This sawtooth function is another example of a discontinuous function. The values of x where the value of f (x) makes a jump are called points of discontinuity of the function f . Thus the points of discontinuity of the sawtooth function are the integers. As x approaches the integer n from the left, the value of f (x) approaches + 12 , but f (x) abruptly jumps to the value − 12 when x = n. A precise definition of continuity and discontinuity for functions appears in Section 2.4. Figure 1.2.15 shows a graphing calculator prepared to graph the sawtooth function. ◗ TEXAS INSTRUMENTS TI-83 y t … … 1 −2 −1 1 2 x −1 FIGURE 1.2.14 The graph of the sawtooth function f (x) = x − [[x]] − 12 of Example 7. 16 FIGURE 1.2.15 A graphing calculator prepared to graph the sawtooth function of Example 7. Graphs of Equations and Functions SECTION 1.2 17 Parabolas The graph of a quadratic function of the form f (x) = ax 2 + bx + c (a = 0) (8) is a parabola whose shape resembles that of the particular parabola in Example 8. EXAMPLE 8 Construct the graph of the parabola y = x 2 . Solution We plot some points in a short table of values. x −3 −2 −1 0 1 2 3 9 4 1 0 1 4 9 y = x2 When we draw a smooth curve through these points, we obtain the curve shown in ◗ Fig. 1.2.16. y (−3, 9) (3, 9) y = x2 (−2, 4) (2, 4) (−1, 1) (1, 1) (0, 0) x FIGURE 1.2.16 The graph of the parabola y = x 2 of Example 8. The parabola y = −x 2 would look similar to the one in Fig. 1.2.16 but would open downward instead of upward. More generally, the graph of the equation y = ax 2 y y= x x y=− x FIGURE 1.2.17 The graph of the parabola x = y 2 of Example 9. (9) is a parabola with its vertex at the origin, provided that a = 0. This parabola opens upward if a > 0 and downward if a < 0. [For the time being, we may regard the vertex of a parabola as the point at which it “changes direction.” The vertex of a parabola of the form y = ax 2 (a = 0) is always at the origin. A precise definition of the vertex of a parabola appears in Chapter 9.] √ x and g(x) = − x. √ Solution After plotting and connecting points satisfying y = ± x, we obtain the Fig. 1.2.17. This parabola opens to the right. parabola y 2 = x shown in √ √ The upper half is the graph of f (x) = x, the lower half is the graph of g(x) = − x. Thus the union of the graphs of these two functions is the graph of the single equation y 2 = x. (Compare this with the circle of Example 3.) More generally, the graph of the equation EXAMPLE 9 Construct the graphs of the functions f (x) = x = by 2 √ (10) is a parabola with its vertex at the origin, provided that b = 0. This parabola opens to ◗ the right if b > 0 (as in Fig. 1.2.17), but to the left if b < 0. 17 18 CHAPTER 1 Functions, Graphs, and Models The size of the coefficient a in Eq. (9) [or of b in Eq. (10)] determines the “width” of the parabola; its sign determines the direction in which the parabola opens. Specifically, the larger a > 0 is, the steeper the curve rises and hence the narrower the parabola is. (See Fig. 1.2.18.) a=5 a=2 a=1 y a = 1/2 v 4 h a = 1/4 y 2 u P v 0 u (h, k) -2 0 x 2 k x FIGURE 1.2.18 Parabolas with different widths. FIGURE 1.2.19 A translated parabola. The parabola in Fig. 1.2.19 has the shape of the “standard parabola” in Example 8, but its vertex is located at the point (h, k). In the indicated uv-coordinate system, the equation of this parabola is v = u 2 , in analogy with Eq. (9) with a = 1. But the uv-coordinates and x y-coordinates are related as follows: u = x − h, v = y − k. Hence the x y-coordinate equation of this parabola is y − k = (x − h)2 . (11) Thus when the parabola y = x 2 is translated h units to the right and k units upward, the equation in (11) of the translated parabola is obtained by replacing x with x − h and y with y − k. This is another instance of the translation principle that we observed in connection with circles. More generally, the graph of any equation of the form y = ax 2 + bx + c (a = 0) (12) can be recognized as a translated parabola by first completing the square in x to obtain an equation of the form y − k = a(x − h)2 . y (13) The graph of this equation is a parabola with its vertex at (h, k). EXAMPLE 10 Determine the shape of the graph of the equation x y = 2x 2 − 4x − 1. (14) Solution If we complete the square in x, Eq. (14) takes the form y = 2(x 2 − 2x + 1) − 3; (1, −3) FIGURE 1.2.20 The parabola y = 2x 2 − 4x − 1 of Example 10. 18 y + 3 = 2(x − 1)2 . Hence the graph of Eq. (14) is the parabola shown in Fig. 1.2.20. It opens upward and ◗ its vertex is at (1, −3). Graphs of Equations and Functions SECTION 1.2 19 Applications of Quadratic Functions In Section 1.1 we saw that a certain type of applied problem may call for us to find the maximum or minimum attained by a certain function f . If the function f is a quadratic function as in Eq. (8), then the graph of y = f (x) is a parabola. In this case the maximum (or minimum) value of f (x) corresponds to the highest (or lowest) point of the parabola. We can therefore find this maximum (or minimum) value graphically—at least approximately—by zooming in on the vertex of the parabola. For instance, recall the animal pen problem of Section 1.1. In Example 9 there we saw that the area A of the pen (see Fig. 1.2.21) is given as a function of its base length x by x $5/ft y $5/ft $5/ft y $1/ft x Wall A(x) = 35 (30x − x 2 ), 0 x 30. (15) Figure 1.2.22 shows the graph y = A(x), and Figs. 1.2.23, 1.2.24, and 1.2.25 show successive magnifications of the region near the high point (vertex) of the parabola. The dashed rectangle in each figure is the viewing window for the next. Figure 1.2.25 makes it seem that the maximum area of the pen is A(15) = 135. It is clear from the figure that the maximum value of A(x) is within 0.001 of A = 135. FIGURE 1.2.21 The animal pen. 200 140 160 136 132 120 y y 80 124 40 0 y = A(x) 128 y = A(x) 120 0 10 20 30 10 12 x 14 16 18 20 x FIGURE 1.2.22 The graph y = A(x). FIGURE 1.2.23 The first zoom. 136 135.01 135.6 135.2 y y 134.8 135 y = A(x) y = A(x) 134.4 134 14 A 150 Highest point (15, 135) Horizontal tangent line 15 x 15.5 16 FIGURE 1.2.24 The second zoom. 134.99 14.9 14.95 15 x 15.05 15.1 FIGURE 1.2.25 The third zoom. We can verify by completing the square as in Example 10 that the maximum value is precisely A(15) = 135: 100 50 14.5 A = 35 (30x − x 2 ) 10 20 A = − 35 (x 2 − 30x) = − 35 (x 2 − 30x + 225 − 225) 30 = − 35 (x 2 − 30x + 225) + 135; x that is, FIGURE 1.2.26 The graph of A(x) = 35 (30x − x 2 ) for 0 x 30. A − 135 = − 35 (x − 15)2 . (16) It follows from Eq. (16) that the graph of Eq. (15) is the parabola shown in Fig. 1.2.26, which opens downward from its vertex (15, 135). This proves that the maximum value of A(x) on the interval [0, 30] is the value A(15) = 135, as both our 19 20 CHAPTER 1 Functions, Graphs, and Models numerical investigations in Section 1.1 and our graphical investigations here suggest. And when we glance at Eq. (16) in the form A(x) = 135 − 35 (x − 15)2 , it’s clear and unarguable that the maximum possible value of 135 − 35 u 2 is 135 when u = x − 15 = 0—that is, when x = 15. The technique of completing the square is quite limited: It can be used to find maximum or minimum values only of quadratic functions. One of the goals in calculus is to develop a more general technique that can be applied to a far wider variety of functions. The basis of this more general technique lies in the following observation. Visual inspection of the graph of A(x) = 35 (30x − x 2 ) in Fig. 1.2.26 suggests that the line tangent to the curve at its highest point is horizontal. If we knew that the tangent line to a graph at its highest point must be horizontal, then our problem would reduce to showing that (15, 135) is the only point of the graph of y = A(x) at which the tangent line is horizontal. But what do we mean by the tangent line to an arbitrary curve? We pursue this question in Section 2.1. The answer will open the door to the possibility of finding the maximum and minimum values of a wide variety of functions. Graphic, Numeric, and Symbolic Viewpoints An equation y = f (x) provides a symbolic description of the function f . A table of values of f (like those in Section 1.1) is a numeric representation of the function, whereas this section deals largely with graphic representations of functions. Interesting applications often involve looking at the same function from at least two of these three viewpoints. EXAMPLE 11 Suppose that a car begins (at time t = 0 hours) in Athens, Georgia (position x = 0 miles) and travels to Atlanta (position x = 60) with a constant speed of 60 mi/h. The car stays in Atlanta for exactly one hour, then returns to Athens, again with a constant speed of 60 mi/h. Describe the car’s “position function” both graphically and symbolically. Solution It’s fairly clear that x = 60t during the 1-hour trip from Athens to Atlanta; for instance, after t = 12 hour the car has traveled halfway, so x = 30 = 12 · 60. During the next hour, 1 t 2, the car’s position is constant, x ≡ 60. And perhaps you can see that during the return trip of the third hour, 2 t 3, the car’s position is given by x = 60 − 60(t − 2) = 180 − 60t x (so that x(2) = 60 and x(3) = 0). Thus the position function x(t) is defined symbolically by x = x(t ) 60 1 2 3 FIGURE 1.2.27 The graph of the position function x(t) in Example 11. 20 t ⎧ ⎪ ⎨60t x(t) = 60 ⎪ ⎩ 180 − 60t if 0 t 1, if 1 < t 2, if 2 < t 3. The domain of this function is the t-interval [0, 3] and its graph is shown in Fig. 1.2.27, where we denote both the function and the dependent variable by the same symbol x (an abuse of notation that’s not uncommon in applications). ◗ Graphs of Equations and Functions SECTION 1.2 21 50 45 40 35 30 P 25 20 15 10 5 0 EXAMPLE 12 During the decade of the 1980s the population P (in thousands) of a small but rapidly growing city was recorded in the following table. P = P(t) Year 1980 1982 1984 1986 1988 1990 t 0 2 4 6 8 10 P 27.00 29.61 32.48 35.62 39.07 42.85 Estimate the population of this city in the year 1987. 0 2 4 6 8 t FIGURE 1.2.28 The population function of Example 12. 10 Solution Figure 1.2.28 shows a graph of the population function P(t) obtained by connecting the six given data points (t, P(t)) with a smooth curve. A careful measurement of the height of the point on this curve at which t = 7 yields the approximate ◗ population P(7) ≈ 37.4 (thousand) of the city in 1987. 1.2 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Parallel lines, if not vertical, have the same slope. The line with equation y = 3x − 5 has slope 3 and y-intercept 5. The graph of the equation (x − 2)2 + (y + 3)2 = 25 is a circle. The graph of the function f is defined to be the graph of the equation y = f (x). If the number a on the x-axis is in the domain of the function f , then the vertical line through a meets the graph of f in exactly one point. The graph of y = |x| has a discontinuity at x = 0. The graph of the “sawtooth function” of Example 7 has a discontinuity at each integral value of x. If a = 0, then the graph of y = ax 2 is a parabola with its vertex at the origin. The graph of y = 2x 2 − 4x − 1 (Example 10) is a parabola opening upward and having its vertex at the point (1, −3). The position formula x(t) in Example 11 is not a function because its rule is expressed in three parts. 1.2 CONCEPTS: QUESTIONS AND DISCUSSION 1. Two general forms of equations of straight lines are reviewed at the beginning of this section. Describe a straight line for which the slope-intercept equation would be the one more convenient to use in writing an equation of the line. Then describe a line for which the point-slope equation would be more convenient. 2. (a) What is the difference between a line that has slope zero and a line that has no slope? If two lines are perpendicular and one of them has slope zero, what is the slope of the other line? (b) Let L 1 and L 2 be two perpendicular lines having slopes m 1 and m 2 , respectively. Theorem 2 in Appendix B asserts that L 1 and L 2 are perpendicular if and only if m 1 m 2 = −1. Is this assertion true in case L 1 is the x-axis and L 2 is the y-axis? Or is there an oversight in the statement of Theorem 2 in Appendix B? 3. (a) Sketch the graph of the equation |x| + |y| = 1. Is this graph the graph of some function? Justify your answer. (b) Repeat part (a), but with the equation |x + y| = 1. 4. (a) Suppose that f is a function such that f (x) > 0 for all real x. Discuss the question of whether the graph of the given equation is the graph of some function. (i) y 2 = f (x); (ii) |y| = f (x); (iii) y = | f (x)|. 21 22 CHAPTER 1 Functions, Graphs, and Models (b) Repeat part (a), but assume that f (x) < 0 for all x. (c) Repeat part (a), but assume that f has both positive and negative values. For instance, sketch the graphs of the equations in (i), (ii), and (iii) if f (x) = x 2 − 1. 5. Newspaper articles often describe or refer to functions (either explicitly or implicitly) but rarely contain equations. Find and discuss examples of numeric and graphic representations of functions in a typical issue of your local newspaper. Also see if you can find a reference to a function that is described verbally but without either a graphic or a numeric representation. 1.2 PROBLEMS In Problems 1 through 10, write an equation of the line L described and sketch its graph. 1. L passes through the origin and the point (2, 3). 2. L is vertical and has x-intercept 7. 3. L is horizontal and passes through (3, −5). 4. L has x-intercept 2 and y-intercept −3. 5. L passes through (2, −3) and (5, 3). 6. L passes through (−1, −4) and has slope 12 . 7. L passes through (4, 2) and has angle of inclination 135◦ . 8. L has slope 6 and y-intercept 7. 9. L passes through (1, 5) and is parallel to the line with equation 2x + y = 10. 30. f (x) = 1 + 2x 2 31. f (x) = x 3 √ 33. f (x) = 4 − x 2 √ 35. f (x) = x 2 − 9 32. f (x) = x 4 √ 34. f (x) = − 9 − x 2 1 x +2 1 39. f (x) = (x − 1)2 37. f (x) = 10. L passes through (−2, 4) and is perpendicular to the line with equation x + 2y = 17. Sketch the translated circles in Problems 11 through 16. Indicate the center and radius of each. 45. f (x) = √ 11. x + y = 4x 12. x + y + 6y = 0 2 2 2 13. x + y + 2x + 2y = 2 2 2 14. x 2 + y 2 + 10x − 20y + 100 = 0 15. 2x 2 + 2y 2 + 2x − 2y = 1 0x <2 29. f (x) = 10 − x 2 1 2x + 3 √ 43. f (x) = 1 − x 2 41. f (x) = 1 2x + 3 1 1−x 1 38. f (x) = 2 x |x| 40. f (x) = x 36. f (x) = 42. f (x) = 1 (2x + 3)2 1 44. f (x) = √ 1−x 46. f (x) = |2x − 2| 47. f (x) = |x| + x 48. f (x) = |x − 3| 49. f (x) = |2x + 5| 50. f (x) = |x| x2 if x < 0, if x 0 16. 9x 2 + 9y 2 − 6x − 12y = 11 Sketch graphs of the functions given in Problems 51 through 56. Indicate any points of discontinuity. Sketch the translated parabolas in Problems 17 through 22. Indicate the vertex of each. 51. f (x) = 0 1 if x < 0, if x 0 52. f (x) = 1 0 if x is an integer, otherwise 17. y = x 2 − 6x + 9 18. y = 16 − x 2 19. y = x 2 + 2x + 4 20. 2y = x 2 − 4x + 8 21. y = 5x 2 + 20x + 23 22. y = x − x 2 The graph of the equation (x − h)2 + (y − k)2 = C is a circle if C > 0, is the single point (h, k) if C = 0, and contains no points if C < 0. (Why?) Identify the graphs of the equations in Problems 23 through 26. If the graph is a circle, give its center and radius. 23. x 2 + y 2 − 6x + 8y = 0 24. x 2 + y 2 − 2x + 2y + 2 = 0 25. x 2 + y 2 + 2x + 6y + 20 = 0 26. 2x 2 + 2y 2 − 2x + 6y + 5 = 0 Sketch the graphs of the functions in Problems 27 through 50. Take into account the domain of definition of each function, and plot points as necessary. 27. f (x) = 2 − 5x, 22 28. f (x) = 2 − 5x, −1 x 1 53. f (x) = [[2x]] 55. f (x) = [[x]] − x x −1 |x − 1| 56. f (x) = [[x]] + [[−x]] + 1 54. f (x) = In Problems 57 through 64, use a graphing calculator or computer to find (by zooming) the highest or lowest (as appropriate) point P on the given parabola. Determine the coordinates of P with two digits to the right of the decimal correct or correctly rounded. Then verify your result by completing the square to find the actual vertex of the parabola. 57. y = 2x 2 − 6x + 7 58. y = 2x 2 − 10x + 11 59. y = 4x 2 − 18x + 22 60. y = 5x 2 − 32x + 49 61. y = −32 + 36x − 8x 2 Graphs of Equations and Functions SECTION 1.2 23 62. y = −53 − 34x − 5x 2 63. y = 3 − 8x − 3x 72. Figure 1.2.32 64. y = −28 + 34x − 9x 2 2 y In Problems 65 through 68, use the method of completing the square to graph the appropriate function and thereby determine the maximum or minimum value requested. 65. If a ball is thrown straight upward with initial velocity 96 ft/s, then its height t seconds later is y = 96t − 16t 2 (ft). Determine the maximum height that the ball attains. 66. Find the maximum possible area of the rectangle described in Problem 40 of Section 1.1. 67. Find the maximum possible value of the product of two positive numbers whose sum is 50. 68. In Problem 42 of Section 1.1, you were asked to express the daily production of a specific oil field as a function P = f (x) of the number x of new oil wells drilled. Construct the graph of f and use it to find the value of x that maximizes P. In Problems 69 through 72 write a symbolic description of the function whose graph is pictured. You may use the greatest integer function of Examples 6 and 7 (if needed). 69. Figure 1.2.29 y 2 (−2, 1) 1 −2 (−1, 0) 1 2 3 x FIGURE 1.2.29 Problem 69. 70. Figure 1.2.30 y (−2, 2) (2, 2) 1 (−3, 0) −2 −1 (5, 0) 1 −1 2 3 5 x 4 FIGURE 1.2.30 Problem 70. 71. Figure 1.2.31 y 3 2 1 −1 1 −1 −2 FIGURE 1.2.31 Problem 71. −4 −3 −2 −1 1 2 3 4 x −1 −2 FIGURE 1.2.32 Problem 72. Each of Problems 73 through 76 describes a trip you made along a straight road connecting two cities 120 miles apart. Sketch the graph of the distance x from your starting point (in miles) as a function of the time t elapsed (in hours). Also describe the function x(t) symbolically. 73. You traveled for one hour at 45 mi/h, then realized you were going to be late, and therefore traveled at 75 mi/h for the next hour. 74. You traveled for one hour at 60 mi/h, stopped for a half hour while a herd of caribou crossed the road, then drove on toward your destination for the next hour at 60 mi/h. 76. You traveled for a half hour at 60 mi/h, suddenly remembered you had left your wallet at home, drove back at 60 mi/h to get it, and finally drove for two hours at 60 mi/h toward your original destination. 77. Suppose that the cost C of printing a pamphlet of at most 100 pages is a linear function of the number p of pages it contains. It costs $1.70 to print a pamphlet with 34 pages, whereas a pamphlet with 79 pages costs $3.05. (a) Express C as a function of p. Use this function to find the cost of printing a pamphlet with 50 pages. (b) Sketch the straight line graph of the function C( p). Tell what the slope and the C-intercept of this line mean—perhaps in terms of the “fixed cost” to set up the press for printing and the “marginal cost” of each additional page printed. −1 3 1 75. You traveled for one hour at 60 mi/h, were suddenly engulfed in a dense fog, and drove back home at 30 mi/h. (2, 3) 3 2 2 x 78. Suppose that the cost C of renting a car for a day is a linear function of the number x of miles you drive that day. On day 1 you drove 207 miles and the cost was $99.45. On day 2 you drove 149 miles and the cost was $79.15. (a) Express C as a function of x. Use this function to find the cost for day 3 if you drove 175 miles. (b) Sketch the straight line graph of the function C(x). Tell what the slope and the C-intercept of this line mean—perhaps in terms of fixed and marginal costs as in Problem 77. 79. For a Federal Express letter weighing at most one pound sent to a certain destination, the charge C is $8.00 for the first 8 ounces plus 80/ c for each additional ounce or fraction thereof. Sketch the graph of this function C of the total number x of ounces, and describe it symbolically in terms of the greatest integer function of Examples 6 and 7. 80. In a certain city, the charge C for a taxi trip of at most 20 miles is $3.00 for the first 2 miles (or fraction thereof), plus 50/ c for each half-mile (or part thereof) up to a total of 10 miles, plus 50/ c for each mile (or part thereof) over 23 24 CHAPTER 1 Functions, Graphs, and Models 10 miles. Sketch the graph of this function C of the number x of miles and describe it symbolically in terms of the greatest integer function of Examples 6 and 7. 81. The volume V (in liters) of a sample of 3 g of carbon dioxide at 27◦ C was measured as a function of its pressure p (in atmospheres) with the results shown in the following table: p 0.25 1.00 2.50 4.00 6.00 V 6.72 1.68 0.67 0.42 0.27 relating the lengths x and y indicated in Fig. 1.2.33. The graph of Eq. (17) is a translated rectangular hyperbola, while the graph of Eq. (18) is a translated parabola (Fig. 1.2.34). You can use a graphing calculator or computer to locate the pertinent point(s) of intersection of these two graphs. 50 −x x 10 Sketch the graph of the function V ( p) and use the graph to estimate the volumes of the gas sample at pressures of 0.5 and 5 atmospheres. 82. The average temperature T (in ◦ F) in Athens, Georgia was measured at two-month intervals, with the results shown in the following table: Date Jul 15 Sep 15 Nov 15 Jan 15 Mar 15 May 15 y 10 FIGURE 1.2.33 The broken tree. (y + 10)2 = 2500 − 100x 80 60 y = 100 x − 10 40 T 79.1 70.2 52.3 43.4 52.2 70.1 Sketch the graph of T as a function of the number of days after July 15. Then use your graph to estimate the average temperature on October 15 and on April 15. 83. A 50-ft tree stands 10 ft from a fence 10 feet high. The tree is suddenly “broken” part of the way up. You are to determine the height of the break so that the tree falls with its trunk barely touching the top of the fence when the tip of the tree strikes the ground on the other side of the fence. The key is the use of simple geometry to derive the equations 100 y= , (17) x − 10 (y + 10)2 = 2500 − 100x 20 y 0 (25, −10) -20 -40 -60 -80 0 5 10 15 x 20 25 30 FIGURE 1.2.34 The hyperbola and parabola in the broken tree investigation. (18) 1.3 POLYNOMIALS AND ALGEBRAIC FUNCTIONS 5 In this section and the next we briefly survey a variety of functions that are used in applications of calculus to describe and model changing phenomena in the world around us. Our viewpoint here is largely graphical. The objective is for you to attain a general understanding of major differences between different types of functions. In later chapters we use calculus to investigate further the graphs presented here. x6 4 x4 3 x2 y 2 1 0 -1 -2 -1 0 x 1 FIGURE 1.3.1 Graphs of power functions of even degree (Example 1). 24 2 Power Functions A function of the form f (x) = x k (where k is a constant) is called a power function. If k = 0 then we have the constant function f (x) ≡ 1. The shape of the graph of a power function with exponent k = n, a positive integer, depends on whether n is even or odd. EXAMPLE 1 The graphs of the even-degree power functions x 2 , x 4 , x 6 , . . . all “cup upward,” as indicated in Fig. 1.3.1. If n > 2 is an even integer then the graph y = x n resembles the parabola y = x 2 , but is flatter near the origin and steeper when |x| > 1. The graphs of the odd-degree power functions x 1 , x 3 , x 5 , . . . all go “from southwest to northeast,” as indicated in Fig. 1.3.2. If n > 3 is an odd integer then the graph y = x n resembles that of y = x 3 , but again is flatter near the origin and steeper when ◗ |x| > 1. Polynomials and Algebraic Functions SECTION 1.3 25 Note that all the power function graphs in Figs. 1.3.1 and 1.3.2 pass through the origin, through the point (1, 1), and either through (−1, 1) or (−1, −1), depending on whether n is even or odd. In either case, x n increases numerically (either positively or negatively) as x does. Would you agree that the notation 4 x5 x3 2 x y 0 x n → +∞ as x → +∞, xn → -2 -4 -2 -1 0 x 1 2 FIGURE 1.3.2 Graphs of power functions of odd degree (Example 1). 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -5 as x → −∞ as x → −∞ if n is even, if n is odd (with the arrow signifying “goes to”) provides a convenient and suggestive description of the general features, when |x| becomes large, of the graphs in Figs. 1.3.1 and 1.3.2? The graph y = x k may have a quite different appearance if the exponent k is not a positive integer. If k is a negative integer—say, k = −m where m is a positive integer—then 1 f (x) = x k = x −m = m , x so in this case the power function is the reciprocal of a function like those in Example 1. Figures 1.3.3 and 1.3.4 show the graphs of y = x−1 0 x FIGURE 1.3.3 y = 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -5 +∞ −∞ y = x −1 = 5 1 . x y = x−2 1 x and FIGURE 1.3.4 y = 5 1 . x2 1 , x2 respectively. Observe that 0 is not in the domain of such a function. Moreover, the reciprocal of a number close to zero is very large in magnitude, which explains the behavior of these graphs near zero: In both graphs, |y| is very large—so the point (x, y) is either very high or very low—when x is close to zero. The graph y = x k may be undefined if x 0 and k is not an integer. In the simplest such case, when k is irrational, we do not attempt to define x k if x < 0, so the graph of x k exists only for x 0. The situation is still more complicated if the exponent k is not an integer. We do not (at present) attempt to define the expression x k if k is irrational—that is, not a quotient of integers. But if k = m/n is rational, with the integers m and n having no common integral factor larger than 1, then we can write x k = x m/n = 0 x y = x −2 = √ n xm , √ and thereby interpret f (x) = x k as a “root function.” If n is odd then n x m is defined for all real x if m is positive and for √ all nonzero values of x if m is negative. But if n is even and m is odd, then the root n x m is not defined for negative x values. The √ typical behavior of√such root functions is illustrated by the graphs of y√= x 1/2 = x and y = x 1/3 = 3 x shown in√Figs. 1.3.5 and 1.3.6. The square root x is defined only for x 0. The cube root 3 x is defined for all x, but observe that its graph appears to be tangent to the y-axis at the origin. 3 3 2.5 2 y= 2 1.5 y y= x 1 0.5 3 x 1 y 0 -1 0 -0.5 -1 -1 -0.5 0 0.5 1 1.5 2 2.5 3 x FIGURE 1.3.5 y = x 1/2 . -2 -3 -5 0 x 5 FIGURE 1.3.6 y = x 1/3 . 25 26 CHAPTER 1 Functions, Graphs, and Models Combinations of Functions Many varied and complicated functions can be assembled out of simple “buildingblock functions.” Here we discuss some of the ways of combining functions to obtain new ones. Suppose that f and g are functions and that c is a fixed real number. The (scalar) multiple c f , the sum f + g, the difference f − g, the product f · g, and the quotient f /g are the new functions with the following formulas: (c f )(x) = c · f (x), (1) ( f + g)(x) = f (x) + g(x), (2) ( f − g)(x) = f (x) − g(x), (3) ( f · g)(x) = f (x) · g(x), f f (x) (x) = . g g(x) and (4) (5) The combinations in Eqs. (2) through (4) are defined for every number x that lies both in the domain of f and in the domain of g. In Eq. (5) we must also require that g(x) = 0. EXAMPLE 2 Let f (x) = x 2 + 1 and g(x) = x − 1. Then: (3 f )(x) = 3(x 2 + 1), ( f + g)(x) = (x 2 + 1) + (x − 1) = x 2 + x, ( f − g)(x) = (x 2 + 1) − (x − 1) = x 2 − x + 2, ( f · g)(x) = (x 2 + 1)(x − 1) = x 3 − x 2 + x − 1, f x2 + 1 (x) = (x = 1). g x −1 and ◗ √ √ EXAMPLE 3 If f (x) = 1 − x for x 1 and g(x) = 1 + x for x −1, then the sum and product of f and g are defined where both f and g are defined. Thus the domain of both √ √ f (x) + g(x) = 1 − x + 1 + x and f (x) · g(x) = √ 1−x √ 1+x = 1 − x2 is the closed interval [−1, 1]. But the domain of the quotient √ f (x) 1−x 1−x =√ = g(x) 1+x 1+x is the half-open interval (−1, 1], because g(−1) = 0. ◗ The results of algebraic operations can sometimes be visualized with the aid of geometric interpretations of the operations. Figures 1.3.7 through 1.3.10 show the results of various operations involving the function f (x) = 20x 2 (x 2 − 1)2 . Adding a constant simply shifts the graph vertically, as in Fig. 1.3.7, which shows y = f (x) + c for c = −2, 0, 2, and 4. Multiplication by a positive constant c expands (if c > 1) or contracts (if 0 < c < 1) the graph in the vertical direction, as in Fig. 1.3.8, which shows y = c f (x) for c = 1, 2, and 3. Figure 1.3.9 shows y = f (x) and the parabola y = 2x 2 , whereas Fig. 1.3.10 shows the graph y = 2x 2 + f (x), obtained by adding the ordinates of the two curves. 26 Polynomials and Algebraic Functions SECTION 1.3 27 8 c=3 9 8 c=2 7 6 y 5 4 3 2 1 c=1 0 -1 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x c=4 6 c=2 4 y c=0 2 0 -2 c = -2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x FIGURE 1.3.7 y = 20x 2 (x 2 − 1)2 + c for c = −2, 0, 2, 4. FIGURE 1.3.8 y = c · 20x 2 (x 2 − 1)2 for c = 1, 2, 3. y = 20x2 (x2 − 1)2 y y = 2x2 + 20x2 (x2 − 1)2 6 6 5 5 4 4 3 y 3 2 2 1 1 0 0 y = 2x2 y = 2x2 -1 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x -1 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 x FIGURE 1.3.9 y = 2x 2 and y = 20x 2 (x 2 − 1)2 . FIGURE 1.3.10 y = 2x 2 and y = 2x 2 + 20x 2 (x 2 − 1)2 . Polynomials A polynomial of degree n is a function of the form p(x) = an x n + an−1 x n−1 + · · · + a2 x 2 + a1 x + a0 (6) where the coefficients a0 , a1 , . . . , an are fixed real numbers and an = 0. Thus an nth-degree polynomial is a sum of constant multiples of the power functions 1, x, x 2, ... , x n−1 , xn. A first-degree polynomial is simply a linear function a1 x + a0 whose graph is a straight line. A second-degree polynomial is a quadratic function whose graph y = a2 x 2 + a1 x + a0 is a parabola (see Section 1.2). Recall that a zero of the function f is a solution of the equation f (x) = 0. Is it obvious to you that the zeros of f (x) are precisely the x-intercepts of the graph y = f (x)? Indeed, a major reason for being interested in the graph of a function is to see the number and approximate locations of its zeros. A key to understanding graphs of higher-degree polynomials is the fundamental theorem of algebra. It states that every nth-degree polynomial has n zeros (possibly complex, possibly repeated). It follows that an nth-degree polynomial has no more than n distinct real zeros. 27 28 CHAPTER 1 Functions, Graphs, and Models EXAMPLE 4 Figures 1.3.11 and 1.3.12 exhibit polynomials that both have the maximum number of real zeros allowed by the fundamental theorem of algebra. But the graphs of power functions in Figs. 1.3.1 and 1.3.2 show that a high-degree polynomial may have only a single real zero. And the quadratic function 4 2 f (x) = x 2 + 4x + 13 = (x + 2)2 + 9 y 0 has no real zeros at all. (Why not?) Figure 1.3.7 includes graphs of sixth-degree polynomials having six, three, or no zeros. Indeed, an nth-degree polynomial can have ◗ any number of zeros from 0 to n if n is even (from 1 to n if n is odd). -2 -4 -4 -2 0 x 2 A polynomial behaves “near infinity”—that is, outside an interval on the x-axis containing its real zeros—in much the same way as a power function of the same degree. If p(x) is a polynomial of odd degree, then y = p(x) goes in opposite (vertical) directions as x goes to −∞ and to +∞ (like the cubic polynomial graph in Fig. 1.3.11). But if p(x) is a polynomial of even degree, then y = p(x) goes in the same (vertical) direction as x goes to −∞ and to +∞ (like the 4th-degree polynomial graph in Fig. 1.3.12). Between the extremes to the left and right, where |x| is large, an nth-degree polynomial has at most n − 1 “bends”—like the 2 bends of the 3rd-degree polynomial graph in Fig. 1.3.11 and the 3 bends of the 4th-degree polynomial graph in Fig. 1.3.12. In Chapter 4 we will use calculus to see why this is so (and to make precise the notion of a “bend” in a curve). 4 FIGURE 1.3.11 f (x) = x 3 − 3x 2 + 1 has three real zeros (Example 4). 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -3 Calculator/Computer Graphing -2 -1 0 x 1 2 3 FIGURE 1.3.12 f (x) = x 4 − 4x 2 + x + 1 has four real zeros (Example 4). A typical calculator or computer graphing utility shows (on its graphics screen or monitor) only that portion of a graph y = f (x) that lies within a selected rectangular viewing window of the form and c y d }. {(x, y) : a x b The parts of the graph that lie outside this viewing window remain unseen (Fig. 1.3.13). With a calculator the maximum and minimum x- and y-values may be entered explicitly in a form such as Xmin = a Xmax = b Ymin = c Ymax = d Frequently the user must specify the x-range [a, b] and the y-range [c, d] carefully so that the viewing window will show the desired portion of the graph. The calculator or computer’s “default window” may provide only a starting point. y d c a b x FIGURE 1.3.13 The viewing window a x b, c y d. EXAMPLE 5 Construct a graph that exhibits the principal features of the cubic polynomial y = x 3 + 12x 2 + 5x − 66. 28 (7) Polynomials and Algebraic Functions SECTION 1.3 29 Solution We anticipate a graph that looks somewhat like the cubic graph in Fig. 1.3.11, one that goes “from southwest to northeast,” perhaps with a couple of bends in between. But when we enter Eq. (7) in a typical graphing calculator with default viewing window −10 x 10, −10 y 10, we get the result shown in Fig. 1.3.14. Evidently our viewing window is not large enough to show the expected behavior. 10 8 6 4 2 y 0 -2 -4 -6 -8 -10 -10 200 15 150 10 100 5 50 y 0 y 0 -5 0 x 5 10 FIGURE 1.3.14 y = x 3 + 12x 2 + 5x − 66 with viewing window −10 x 10, −10 y 10. 1000 800 600 400 200 y 0 -200 -400 -600 -800 -1000 -5 20 -5 -50 -10 -100 -15 -150 -20 -20 -15 -10 -5 0 x 5 10 15 20 FIGURE 1.3.15 y = x 3 + 12x 2 + 5x − 66 with viewing window −20 x 20, −20 y 20. -200 -20 -15 -10 -5 0 x 5 10 15 20 FIGURE 1.3.16 y = x 3 + 12x 2 + 5x − 66 with viewing window −20 x 20, −200 y 200. Doubling each dimension of the viewing window, we get the result in Fig. 1.3.15. Now we see the three zeros that a cubic polynomial can have, as well as some possibility of two bends, but it appears that magnification in the y-direction is indicated. Perhaps we need a y-range measuring in the hundreds rather than the tens. With the viewing window −20 x 20, −200 y 200 we finally get the satisfying graph shown in Fig. 1.3.16. Once we have zoomed out to see the “big picture,” we can zoom in on points of interest. For instance, Fig. 1.3.16 indicates “zoom boxes” locating the three zeros of the polynomial in (7). Apparently these zeros are located at or near the points x = −11, x = −3, and x = 2. Each can be approximated graphically as closely as you please (subject to the limitations of your computer) by the method of successive magnifications. (See if you can convince yourself that these three zeros are exactly the ◗ indicated integers. How could you verify that this actually is true?) 0 5 x 10 15 FIGURE 1.3.17 y = (x 2 − 1)(x − 10)(x − 10.1) with viewing window −5 x 15, −1000 y 1000. 1 0.5 y 0 EXAMPLE 6 Investigate the graph of the quartic (fourth-degree) polynomial f (x) = (x 2 − 1)(x − 10)(x − 10.1) = x 4 − (20.1)x 3 + 100x 2 + (20.1)x − 101. (8) Solution Here we know the zeros x = −1, 1, 10, and 10.1 in advance, so it makes sense to choose an x-range that includes all four. Noting that f (0) = −101, we suspect that a y-range measuring in the hundreds is indicated. Thus with the viewing window −5 x 15, −1000 y 1000, we get the attractive graph in Fig. 1.3.17. Observe that with its three bends it resembles the quartic graph in Fig. 1.3.12. But now the behavior of the graph near the point x = 10 is unclear. Does it dip beneath the x-axis or not? We select the viewing window 9.5 x 10.5, −1 y 1 to magnify this area and get the result in Fig. 1.3.18. This is a case where it appears that different plots on different scales are required to show all the behavior of the graph. ◗ -0.5 -1 9.5 10 x 10.5 FIGURE 1.3.18 y = (x 2 − 1)(x − 10)(x − 10.1) with viewing window 9.5 x 10.5, −1 y 1. Our graphs in Examples 5 and 6 exhibit the maximum possible number of zeros and bends for the polynomials in Eqs. (7) and (8), so we are fairly confident that our investigations reveal the main qualitative features of the graphs of these polynomials. But only with the calculus techniques of Chapter 4 can we be certain of the structure of a graph. For instance, a polynomial graph can exhibit fewer than the maximum possible number of bends, but at this stage we cannot be certain that more bends are not hidden somewhere, perhaps visible only on a scale different from that of the viewing window we have selected. 29 30 CHAPTER 1 Functions, Graphs, and Models Rational Functions Just as a rational number is a quotient of two integers, a rational function is a quotient f (x) = p(x) q(x) (9) of two polynomials p(x) and q(x). Graphs of rational functions and polynomials have several features in common. For instance, a rational function has only a finite number of zeros, because f (x) in Eq. (9) can be zero only when the numerator polynomial p(x) is zero. Similarly, the graph of a rational function has only a finite number of bends. But the denominator polynomial q(x) in Eq. (9) may have a zero at a point x = a where the numerator is nonzero. In this case the value of f (x) will be very large in magnitude when x is close to a. This observation implies that the graph of a rational function may have a feature that no polynomial graph can have—an asymptote. EXAMPLE 7 Figure 1.3.19 shows the graph of the rational function f (x) = (x + 2)(x − 1) . x(x + 1)(x − 2) (10) Note the x-intercepts x = −2 and x = 1, corresponding to the zeros of the numerator (x + 2)(x − 1). The vertical lines x = −1, x = 0, and x = 2 shown in the graph correspond to the zeros of the denominator x(x + 1)(x − 2). These vertical lines are asymptotes of the graph of f . ◗ 8 8 4 4 y 0 y 0 -4 -4 -8 -8 -4 -2 0 x 2 4 FIGURE 1.3.19 The graph of the rational function in Eq. (10) (Example 7). -4 -2 0 x 2 4 FIGURE 1.3.20 The graph of the rational function in Eq. (11) (Example 8). EXAMPLE 8 Figure 1.3.20 shows the graph of the rational function f (x) = x(x + 2)(x − 1) . (x + 1)(x − 2) (11) The x-intercepts x = −2, x = 0, and x = 1 correspond to the zeros of the numerator, whereas the asymptotes x = −1 and x = 2 correspond to the zeros of the denominator. ◗ It should be clear that—by counting x-intercepts and asymptotes—you could match the rational functions in Eqs. (10) and (11) with their graphs in Figs. 1.3.19 and 1.3.20 without knowing in advance which was which. 30 Polynomials and Algebraic Functions SECTION 1.3 31 3 Algebraic Functions 2.5 An algebraic function is one whose formula can be constructed beginning with power functions and applying the algebraic operations of addition, subtraction, multiplication by a real number, multiplication, division, and/or root-taking. Thus polynomials and rational functions are algebraic functions. But whereas every polynomial is defined everywhere on the real line, and every rational function is defined everywhere except at the (finitely many) real zeros of its denominator (which correspond to vertical asymptotes), the domain of definition of an algebraic function may be quite limited. For instance, Figs. 1.3.21 and 1.3.22 show the graphs of the algebraic functions 2 1.5 y 1 0.5 0 -0.5 -1 -3 -2 -1 0 x 1 2 3 √ 4 FIGURE 1.3.21 y = 16 − x 4 on [−2, 2]. 30 25 20 15 y 10 5 0 -5 -10 -20 -15 -10 -5 0 x FIGURE 1.3.22 y = (−∞, −4] ∪ [4, ∞). 5 10 15 20 √ x 2 − 16 on f (x) = 4 16 − x 4 g(x) = and x 2 − 16 on the bounded and unbounded intervals (respectively) where they are defined. The graph of every polynomial or rational function looks “smooth” at every point where it is defined, but the graph of an algebraic function may exhibit “corners” or sharp “cusps” where it does not look smooth. For instance, look at the graphs in Figs. 1.3.23 and 1.3.24 of the algebraic functions √ 3 f (x) = x 2 = |x| and g(x) = x 2 (x − 2)2 . In Chapter 3 we will use concepts of calculus to say precisely what is meant by a smooth graph. Figure 1.3.25 shows the graphs of the two algebraic functions defined by √ (12) y = ± 0.2969 x − 0.126x − 0.3516x 2 + 0.2843x 3 − 0.10151x 4 . The loop describes the cross-sectional profile of the NASA 0012 airfoil as designed by aeronautical engineers. 5 5 4 4 3 3 y 2 y 2 1 1 0 0 1 0.5 y 0 -0.5 -1 -3 -2 -1 0 x 1 2 FIGURE 1.3.23 y = |x| with a “corner” at the origin. 3 -1 -3 -2 -1 0 1 x 2 3 4 5 FIGURE 1.3.24 y = 3 x 2 (x − 2)2 with “cusps” at (0, 0) and (2, 0). -1 0 0.5 1 x 1.5 2 FIGURE 1.3.25 y= √ ±(0.2969 x − 0.126x − 0.3516x 2 + 0.2843x 3 − 0.10151x 4 ). 1.3 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If x is close to zero, then so is x −3 . √ 2. If m and n are positive integers and x 0, then x m/n = n x m . 3. The product f · g of the functions f and g is defined as follows: ( f · g)(x) = f (x) · g(x). √ √ 4. If f (x) = 1 − x and g(x) = 1 + x, then the domain of f /g is [−1, 1]. 5. If p(x) = x 3 + x 3/2 − x 2 + 1, then p(x) is a polynomial. 6. The quotient of any two functions is known as a rational function. √ 7. If f (x) = |x|, then f is an algebraic function because f (x) = x 2 . 31 32 CHAPTER 1 Functions, Graphs, and Models 8. The graph of the rational function f (x) = x(x + 2)(x − 1) (x + 1)(x − 2) has three x-intercepts and two vertical asymptotes. 9. The graph shown in Fig. 1.3.25 is not the graph of a function. 10. If p(x) is a polynomial of high degree, then as x → +∞, either p(x) → +∞ or p(x) → −∞. 1.3 CONCEPTS: QUESTIONS AND DISCUSSION 1. In each of the following eight cases, give an example of a function as described or explain why no such function exists. (a) A polynomial function of degree less than 2 whose graph lies entirely above the x-axis. (b) A polynomial of positive degree whose graph lies entirely beneath the x-axis. (c) A polynomial of positive degree and with positive leading coefficient whose graph lies entirely below the x-axis (the leading coefficient of a polynomial is the coefficient of its term of highest degree). (d) A polynomial of odd degree with negative leading coefficient whose graph does not intersect the x-axis. (e) A polynomial whose graph lies entirely between the lines y = −1 and y = 1. (f) A polynomial whose graph contains points above the line y = 1 and below the line y = −1, but contains no points between those two lines. (g) A rational function that has both positive and negative values but is never zero. (h) A nonconstant rational function that is never zero and has no vertical asymptote. 2. In each of the following five cases write the formula of a specific function as described. Also sketch a typical graph of such a function (not necessarily the same one you defined symbolically). (a) (b) (c) (d) (e) A quadratic polynomial with no real zeros. A cubic polynomial with exactly one real zero x = 0. A cubic polynomial with exactly two distinct real zeros. A quartic polynomial with exactly two distinct real zeros. A quartic polynomial with exactly three distinct real zeros. 3. Which of the following algebraic functions agrees with some polynomial function? √ √ (b) f (x) = x 4 + 4x + 4 (a) f (x) = x 2 + 2x + 1 (c) f (x) = 3 (x − 1)3 (d) f (x) = 1.3 PROBLEMS In Problems 1 through 6, find f + g, f · g, and f /g, and give the domain of definition of each of these new functions. 1. f (x) = x + 1, g(x) = x 2 + 2x − 3 1 1 2. f (x) = , g(x) = x −1 2x + 1 √ √ 3. f (x) = x, g(x) = x − 2 32 4. f (x) = 5. f (x) = 6. f (x) = √ √ x + 1, x 2 + 1, x −1 , x −2 3 (x − 2)2 g(x) = √ 5−x g(x) = √ g(x) = 1 4 − x2 x +1 x +2 Polynomials and Algebraic Functions SECTION 1.3 33 In Problems 7 through 12, match the given polynomial with its graph among those shown in Figs. 1.3.26 through 1.3.31. Do not use a graphing calculator or a computer. Instead, consider the degree of the polynomial, its indicated number of zeros, and its behavior for |x| large. 4 4 2 2 y 0 y 0 -2 -2 7. f (x) = x 3 − 3x + 1 8. f (x) = 1 + 4x − x -4 -4 -4 3 9. f (x) = x 4 − 5x 3 + 13x + 1 0 x -4 4 FIGURE 1.3.32 11. f (x) = 16 + 2x 2 − x 4 4 2 2 y 0 y 0 -2 -2 12. f (x) = x 5 + x -4 20 0 x 2 4 2 4 FIGURE 1.3.33 4 10. f (x) = 2x 5 − 10x 3 + 6x − 1 -2 -4 20 -4 10 10 y 0 y 0 -10 -10 -20 -2 -1 0 x 1 2 FIGURE 1.3.26 -20 -3 0 x 2 4 -4 FIGURE 1.3.34 -1.5 0 x 1.5 3 FIGURE 1.3.27 4 -2 -2 0 x FIGURE 1.3.35 In Problems 17 through 20, use primarily the domain of definition of the given algebraic function (rather than a graphing calculator or computer) to match it with its graph among those in Figs. 1.3.36 through 1.3.39. √ √ 17. f (x) = x x + 2 18. f (x) = 2x − x 2 √ √ 19. f (x) = x 2 − 2x 20. f (x) = 2 3 x 2 − 2x 3 4 2 2 10 2 y 0 y 0 y 1 -2 y 0 0 -10 -4 -4 -2 0 x 2 4 FIGURE 1.3.28 -2 -4 -2 0 x 2 -1 -1 4 FIGURE 1.3.29 0 1 x 2 3 -2 0 2 4 x FIGURE 1.3.36 FIGURE 1.3.37 4 6 20 4 4 2 10 2 y y 0 y 0 y 2 0 0 -2 -10 -2 -4 -4 -2 0 x 2 4 -20 -4 0 x 4 -2 0 x 2 FIGURE 1.3.38 FIGURE 1.3.30 1 (x + 1)(x − 2) 3 15. f (x) = 2 x +1 -2 0 x 2 4 FIGURE 1.3.39 FIGURE 1.3.31 In Problems 13 through 16, use the vertical asymptotes of the given rational function (rather than a graphing calculator or computer) to match it with its graph among those shown in Figs. 1.3.32 through 1.3.35. 13. f (x) = -2 -4 14. f (x) = x2 x −9 x2 + 1 16. f (x) = 3 x −1 In Problems 21 through 30 use a graphing calculator or computer to determine one or more appropriate viewing windows to exhibit the principal features of the graph y = f (x). In particular, determine thereby the number of real solutions of the equation f (x) = 0 and the approximate location (to the nearest integer) of each of these solutions. 21. f (x) = x 3 − 3x + 1 22. f (x) = x 3 − 3x + 2 23. f (x) = x 3 − 3x + 3 33 34 CHAPTER 1 Functions, Graphs, and Models 24. f (x) = 2x 4 − 6x 3 + 10x − 5 38. Use the graphical method of repeated magnifications to find both the length and the maximum width of the airfoil shown in Fig. 1.3.25. Determine each accurate to three decimal places. 25. f (x) = 2x − 6x + 10x − 6 4 3 26. f (x) = 2x 4 − 6x 3 + 10x − 7 27. f (x) = x 3 − 50x − 100 39. A 12-ft ladder leans across a 5-ft fence and touches a high wall located 3 ft behind the fence. You are to find the distance from the foot of the ladder to the bottom of the fence. The key is the use of simple geometry to derive the equations 28. f (x) = x 4 + 20x 3 − 50x − 30 29. f (x) = x 5 + 5x 4 − 100x 3 − 200x 2 + 2500x − 3500 30. f (x) = x 6 − 250x 4 + 2500x 2 − 2500 x y = 15 In Problems 31 through 37, determine how the graph y = f (x) changes when the value of c is changed within the given interval. With a graphing calculator or computer you should be able to plot graphs with different values of c on the same screen. (x + 3)2 + (y + 5)2 = 144 and relating the lengths x and y indicated in Fig. 1.3.40. Can you eliminate y to find a quartic polynomial equation that x must satisfy? If so, then you can use a graphing calculator or computer to approximate the possible values of x by the method of repeated magnification. 31. f (x) = x 3 − 3x + c, −5 c 5 32. f (x) = x 3 + cx, −5 c 5 33. f (x) = x 3 + cx 2 , −5 c 5 y 34. f (x) = x 4 + cx 2 , −5 c 5 12 35. f (x) = x 5 + cx 3 + x, −5 c 5 1 , 1 c 10 36. f (x) = 1 + cx 2 x2 37. f (x) = , 1 c 10, x in (−c, c) 2 c − x2 5 x 3 Ground FIGURE 1.3.40 The leaning ladder. 1.4 TRANSCENDENTAL FUNCTIONS Continuing the survey of elementary functions begun in Section 1.3, we now review briefly the most familiar nonalgebraic functions that are studied in calculus. These include the trigonometric functions that are used to model periodic phenomena— phenomena of ebb and flow, involving quantities that oscillate with the passage of time—and the exponential and logarithmic functions that are used to model phenomena of growth and decay—involving quantities that either increase steadily or decrease steadily as time passes. We also introduce composition of functions, a new way (in addition to the algebraic operations of Section 1.3) of combining familiar functions to form new ones. Trigonometric Functions A review of trigonometry is included in Appendix C. In elementary trigonometry a trigonometric function such as sin A, cos A, or tan A ordinarily is first defined as a function of an angle A in a right triangle. But here a trigonometric function of a real number x corresponds to that function of an angle measuring x radians. Thus π √ sin 1 1 π π 3 π 6 =√ , and tan = sin = , cos = π 6 2 6 2 6 3 cos 6 because π/6 is the radian measure of an angle of 30◦ . Recall that π radians = 180 degrees, so 1 rad = 34 180 deg π and 1 deg = π rad. 180 (1) Transcendental Functions SECTION 1.4 35 2 2 y y = cos x y = sinx y=1 1 y 0 -1 -2 0 5 0 -1 y = −1 -5 y=1 1 10 -2 15 y = −1 -5 0 5 x x FIGURE 1.4.1 y = sin x. FIGURE 1.4.2 y = cos x. 10 15 Figures 1.4.1 and 1.4.2 show the graphs y = sin x and y = cos x of the sine and cosine functions, respectively. The value of each oscillates between +1 and −1, exhibiting the characteristic periodicity of the trigonometric functions: sin(x + 2π ) = sin x cos(x + 2π) = cos x and (2) for all x. If we translate the graph y = cos x by π/2 units to the right, we get the graph y = sin x. This observation corresponds to the familiar relation cos x − π 2 = cos π − x = sin x. 2 (3) EXAMPLE 1 Figure 1.4.3 shows the translated sine curve obtained by translating the origin to the point (1, 2). Its equation is obtained upon replacing x and y in y = sin x with x − 1 and y − 2, respectively: y − 2 = sin(x − 1); that is, y = 2 + sin(x − 1). ◗ 4 3 2 (1, 2) y 1 0 -1 T 80 4 8 0 4 8 FIGURE 1.4.3 The translated sine curve y − 2 = sin(x − 1). 60 0 -4 x πt T = 61.3 + (17.9) cos 6 12 t FIGURE 1.4.4 Average daily temperature in Athens, Georgia, t months after July 15 (Example 2). The world around us is full of quantities that oscillate like the trigonometric functions. Think of the alternation of day and night, the endless repetition of the seasons, the monthly cycle of the moon, the rise and fall of the tides, the beat of your heart. EXAMPLE 2 Figure 1.4.4 shows the cosine-like behavior of temperatures in Athens, Georgia. The average temperature T (in ◦ F) during a 24-hr day t months after July 15 is given approximately by T = T (t) = 61.3 + 17.9 cos πt . 6 (4) 35 36 CHAPTER 1 Functions, Graphs, and Models For instance, on a typical October 15 (three months after July 15) the average temperature is 3π = 61.3 (◦ F) T (3) = 61.3 + 17.9 cos 6 because cos(3π/6) = cos(π/2) = 0. Thus the “midpoint” of fall weather in Athens— when the average daily temperature is midway between summer’s high and winter’s low—occurs about three weeks after the official beginning of fall (on or about September 22). Note also that πt πt + 2π = 61.3 + 17.9 cos = T (t) T (t + 12) = 61.3 + 17.9 cos 6 6 (why?), in agreement with the yearly 12-month cycle of average weather. ◗ The periodicity and oscillatory behavior of the trigonometric functions make them quite unlike polynomial functions. Because sin nπ = 0 cos(2n + 1) and π =0 2 (5) for n = 0, 1, 2, 3, . . . , we see that the simple trigonometric equations sin x = 0 cos x = 0 and (6) have infinitely many solutions. In contrast, a polynomial equation can have only a finite number of solutions. Figure 1.4.5 shows the graph of y = tan x. The x-intercepts correspond to the zeros of the numerator sin x in the relation tan x = sin x , cos x (7) whereas the vertical asymptotes correspond to the zeros of the denominator cos x. Observe the “infinite gaps” in the graph y = tan x at these odd-integral multiples of π/2. We call these gaps discontinuities, phenomena we discuss further in Chapter 2. 4 2 y 0 π -π 2π -2 -4 -4 -2 0 2 4 6 x FIGURE 1.4.5 y = tan x. Composition of Functions Many varied and complex functions can be “put together” by using quite simple “building-block” functions. In addition to adding, subtracting, multiplying, or dividing two given functions, we can also combine functions by letting one function act on the output of the other. 36 Transcendental Functions SECTION 1.4 37 DEFINITION Composition of Functions The composition of the two functions f and g is the function h = f ◦ g defined by h(x) = f (g(x)) (8) for all x in the domain of g such that u = g(x) is in the domain of f . (The righthand side in Eq. (8) is read “ f of g of x.”) x Thus the output u = g(x) of the function g is used as the input to the function f (Fig. 1.4.6). We sometimes refer to g as the inner function and to f as the outer function in Eq. (8). g u = g(x) EXAMPLE 3 If f (x) = √ x and g(x) = 1 − x 2 , then 1 − x2 f (g(x)) = f whereas f(u) = f(g(x)) = h(x) g( f (x)) = 1 − FIGURE 1.4.6 The composition of f and g. for |x| 1, √ 2 x =1−x for x 0. ◗ The f (g(x)) notation for compositions is most commonly used in ordinary computations, whereas the f ◦ g notation emphasizes that the composition may be regarded as a new kind of combination of the functions f and g. But Example 3 shows that f ◦ g is quite unlike the product f g of the two functions f and g, for f ◦ g = g ◦ f, whereas f g = g f (because f (x) · g(x) = g(x) · f (x) whenever f (x) and g(x) are defined). So remember that composition is quite different in character from ordinary multiplication of functions. EXAMPLE 4 If f (x) = x 2 and g(x) = cos x, then the functions f (x)g(x) = x 2 cos x, f (g(x)) = cos2 x = (cos x)2 , and g( f (x)) = cos x = cos(x ) 2 2 are defined for all x. Figures 1.4.7 through 1.4.9 illustrate vividly how different these ◗ three functions are. EXAMPLE 5 Given the function h(x) = (x 2 + 4)3/2 , find two functions f and g such that h(x) = f (g(x)). 2 2 200 y = x2 cos x 100 y = cos2 x 1 0 y 0 y 0 -100 -1 -1 y -200 -10 0 x FIGURE 1.4.7 y = x 2 cos x (Example 4). 10 -2 -10 0 x FIGURE 1.4.8 y = cos2 x (Example 4). 10 y = cos x2 1 -2 -10 -5 0 x 5 10 FIGURE 1.4.9 y = cos x 2 (Example 4). 37 38 CHAPTER 1 Functions, Graphs, and Models Solution It is technically correct—but useless—simply to let g(x) = x and f (u) = (u 2 + 4)3/2 . We seek a nontrivial answer here. To calculate (x 2 + 4)3/2 , we must first calculate x 2 + 4. So we choose g(x) = x 2 + 4 as the inner function. The last step is to raise u = g(x) to the power 32 , so we take f (u) = u 3/2 as the outer function. Thus if f (x) = x 3/2 and g(x) = x 2 + 4, then f (g(x)) = f (x 2 + 4) = (x 2 + 4)3/2 = h(x). ◗ Exponential Functions An exponential function is a function of the form f (x) = a x , y = 10x 8 y 4 y = 2x (0, 1) 0 -2 0 2 x FIGURE 1.4.10 Increasing exponential functions y = 2x and y = 10x . 4 (9) where the base a is a fixed positive real number—a constant. Note the difference between an exponential function and a power function. In the power function x n , the variable x is raised to a constant power; in the exponential function a x , a constant is raised to a variable power. Many computers and programmable calculators use the notation a ∧ x to denote the exponential a x (a few use a ↑ x). If a > 1, then the graph y = a x looks much like those in Fig. 1.4.10, which shows y = 2x and y = 10x . The graph of an exponential function with base a, a > 1, lies entirely above the x-axis and rises steadily from left to right. Therefore, such a graph is nothing like the graph of a polynomial or trigonometric function. The larger the base a, the more rapid the rate at which the curve y = a x rises (for x > 0). Thus y = 10x climbs more steeply than y = 2x . EXAMPLE 6 Every exponential function (with base a > 1) increases very rapidly when x is large. The following table comparing values of x 2 with 2x exhibits vividly the rapid rate of increase of the exponential function 2x , even compared with the power function x 2 , which increases at a more restrained rate as x increases. x x2 2x 10 20 30 40 50 60 70 80 90 100 100 400 900 1600 2500 3600 4900 6400 8100 10000 1024 1048576 1073741824 1099511627776 1125899906842624 1152921504606846976 1180591620717411303424 1208925819614629174706176 1237940039285380274899124224 1267650600228229401496703205376 The comparison between x 2 and 2x for smaller values of x is interesting in a different way. The graphs of y = x 2 and y = 2x in Fig. 1.4.11 indicate that the equation x 2 = 2x has three solutions between x = −2 and x = 5. Is it clear to you that x = 2 and x = 4 are exact solutions? The “zoom” shown in Fig. 1.4.12 indicates that the negative solution is a bit less than −0.75. Perhaps you can zoom once more and find ◗ the value of this negative solution accurate to at least two decimal places. If we replace x in Eq. (9) with −x, we get the function a −x . Its graph y = a −x falls from left to right if a > 1. Figure 1.4.13 shows the graphs y = 3−x and y = 7−x . Whereas trigonometric functions are used to describe periodic phenomena of ebb and flow, exponential functions are used to describe natural processes of steady growth or steady decline. 38 Transcendental Functions SECTION 1.4 39 25 20 (4, 16) 15 y = x2 y 10 y = 2x 5 (?, ?) (2, 4) 0 -5 -2 -1 0 1 2 3 4 5 1 0.9 0.8 0.7 0.6 y 0.5 0.4 0.3 0.2 0.1 0 -1 y=7-x 8 y = 2x y 4 y = 3-x y = x2 0 -0.5 x x FIGURE 1.4.11 y = x 2 and y = 2x . 0 FIGURE 1.4.12 A magnification of 1.4.11 showing the negative solution. -2 0 2 4 x FIGURE 1.4.13 Decreasing exponential functions y = 3−x and y = 7−x . EXAMPLE 7 Let P(t) denote the number of rodents after t months in a certain prolific population that doubles every month. If there are P(0) = 10 rodents initially, then there are • • • P(1) = 10 · 21 = 20 rodents after 1 month, P(2) = 10 · 22 = 40 rodents after 2 months, P(3) = 10 · 23 = 80 rodents after 3 months, and so forth. Thus the rodent population after t months is given by the exponential function (10) P(t) = 10 · 2t if t is a nonnegative integer. Under appropriate conditions, Eq. (10) gives an accurate approximation to the rodent population even when t is not an integer. For instance, this formula predicts that after t = 4 12 months, there will be P(4.5) = 10 · 24.5 ≈ 226.27 ≈ 226 rodents. ◗ EXAMPLE 8 Suppose that you invest $5000 in a money-market account that pays 8% interest compounded annually. This means that the amount in the account is multiplied by 1.08 at the end of each year. Let A(t) denote the amount in your account at the end of t years. Then, • • 12000 • A = 10,000 ($5400.00) after 1 yr, ($5832.00) after 2 yr, ($6298.56) after 3 yr, and so on. Thus after t years (t a nonnegative integer), the amount in your account is given by the exponential function A 8000 A = 5000 . 1.08 A(1) = 5000 · 1.081 A(2) = 5000 · 1.082 A(3) = 5000 · 1.083 t A(t) = 5000 · 1.08t . 4000 0 0 2 4 6 8 t FIGURE 1.4.14 The graph for Example 8. 10 (11) Figure 1.4.14 shows the graph A(t) = 5000 · 1.08t as well as the horizontal line A = 10,000. From this graph we see, for instance, that the amount in the account has doubled (to $10,000) after approximately t = 9 yr. We could approximate the “doubling time” more accurately by magnifying the graph near the intersection of the ◗ horizontal line and the rising curve. Example 9 exhibits a function that combines the steady decrease of an exponential function with negative exponent with the oscillation of a trigonometric function. EXAMPLE 9 The function y(t) = 3 · 2−t cos 4π t (12) 39 40 CHAPTER 1 Functions, Graphs, and Models 3 might describe the amplitude y, in inches, of the up-and-down vibrations of a car with very poor shock absorbers t seconds after it hits a deep pothole. Can you see that Eq. (12) describes an initial (t = 0) amplitude of 3 inches that halves every second, while two complete up-and-down oscillations occur every second? (The factor 3 · 2−t is the decreasing amplitude of the vibrations, while the function cos 4π t has period 12 s.) Figure 1.4.15 shows the graph of y(t). The curve described in Eq. (12) oscillates between the two curves y(t) = ±3 · 2−t . It appears that the car’s vibrations subside ◗ and are negligible after 7 or 8 seconds. y = 3 · 2−t 2 y = 3 · 2−t cos 4πt 1 y in. 0 -1 y = −3 · 2−t -2 -3 0 1 2 3 4 5 t sec 6 7 8 Logarithmic Functions In analogy with the inverse trigonometric functions that you may have seen in trigonometry, logarithms are “inverse” to exponential functions. The base a logarithm of the positive number x is the power to which a must be raised to get x. That is, FIGURE 1.4.15 y(t) = 3 · 2−t cos 4π t (Example 9). y = loga x 4 y = ln x y = log10 x 2 y 0 (1, 0) ln x = loge x, where e is a special irrational number: -4 0 2 4 6 8 e = 2.71828182845904523536 . . . . 10 x FIGURE 1.4.16 The common and natural logarithm functions. 6 5 y = log x 4 y = 1 x1/5 2 y 3 You’ll see the significance of this strange-looking base in Chapter 3. Figure 1.4.16 shows the graphs y = ln x and y = log10 x. Both graphs pass through the point (1, 0) and rise steadily (though slowly) from left to right. Because exponential functions never take on zero or negative values, neither zero nor any negative number is in the domain of any logarithmic function. The facts that log10 100,000 = 5 and log10 1,000,000 = 6 indicate that the function log x = log10 x increases quite slowly as x increases. Whereas Example 6 above illustrates the fact that an exponential function a x (with a > 1) increases more rapidly than any power function as x → ∞, Example 10 illustrates the fact that a logarithmic function increases more slowly than any power function. EXAMPLE 10 In the following table we compare the rate of growth of the power function f (x) = 12 x 1/5 with that of the logarithm function g(x) = log x. 2 1 x 0 0.5 1 x 1.5 1 1/5 2x FIGURE 1.4.17 2 × 105 passes log x. 1.5 y = 1 x1/5 2 1 0.5 y 0 y = log x -0.5 -1 -1.5 0 2 4 6 8 10 x FIGURE 1.4.18 log x passes 12 x 1/5 . 40 (13) The LOG key on most calculators gives the base 10 (common) logarithm log10 x. The LN key gives the natural logarithm -2 0 if a y = x. 20000 40000 60000 80000 100000 120000 140000 160000 180000 200000 f (x) = 12 x 1/5 g(x) = log x 3.62390 4.16277 4.51440 4.78176 5 5.18569 5.34805 5.49280 5.62373 5.74349 4.30103 4.60206 4.77815 4.90309 5 5.07918 5.14613 5.20412 5.25527 5.30103 It appears here and in Fig. 1.4.17 that log x is smaller than 12 x 1/5 when x > 100,000. Figure 1.4.18 shows that log x initially is smaller than 12 x 1/5 , but “catches up and passes” 12 x 1/5 somewhere around (although a bit less than) x = 5. Then 12 x 1/5 in turn catches up and passes log x at x = 100,000. When x = 1050 , 12 x 1/5 = 5,000,000,000, ◗ but the value of log x is only 50. Transcendental Functions SECTION 1.4 41 Transcendental Equations The trigonometric, exponential, and logarithmic functions are called transcendental functions. As we saw in Eqs. (5) and (6), an equation that includes transcendental functions can have infinitely many solutions. But it also may have only a finite number of solutions. Determining whether the number of solutions is finite or infinite can be difficult. One approach is to write the given equation in the form f (x) = g(x), (14) where both the functions f and g are readily graphed. Then the real solutions of Eq. (14) correspond to the intersections of the two graphs y = f (x) and y = g(x). EXAMPLE 11 The single point of intersection of the graphs y = x and y = cos x, shown in Fig. 1.4.19, indicates that the equation x = cos x has only a single solution. Moreover, from the graph you can glean the additional information that the solution lies in the interval (0, 1). ◗ 2 y=1−x y=x 4 1 y = 3 cos x y = cos x y 0 y 0 -1 -4 -2 -4 -2 0 x 2 -5 4 FIGURE 1.4.19 Solving the equation x = cos x of Example 11. 0 x 5 FIGURE 1.4.20 Solving the equation 1 − x = 3 cos x of Example 12. EXAMPLE 12 The graphs of y = 1 − x and y = 3 cos x are shown in Fig. 1.4.20. In contrast with Example 11, there are three points of intersection of the graphs. This makes it clear that the equation 1 − x = 3 cos x has one negative solution and two positive solutions. They could be approximated by ◗ (separately) zooming in on the three intersection points. Can You Believe What You See on Your Calculator/Computer Screen? The examples we give next show that the short answer to this question is “not always.” One reason is that a typical graphing calculator or simple computer program plots only a finite number of equally spaced points on the curve y = f (x), a x b, joining the selected points with straight line segments. If the plotted points are sufficiently close, then the resulting graph may look to the unaided eye like a smooth curve, but it may miss some essential features that would be revealed if more points were plotted. EXAMPLE 13 A 1-ampere alternating current with frequency 60 Hz (Hertz; cycles per second) is described by the function I (t) = sin 120πt. (15) The absolute value |I (t)| gives the magnitude (in amperes) of the current at time t, which flows in one direction when I > 0 the opposite direction when I < 0. A simple 41 42 CHAPTER 1 Functions, Graphs, and Models I 1 1 1 0.5 0.5 0.5 I 0 I 0 0 -0.5 -0.5 -0.5 -1 -1 -1 -1 -0.5 0 t 0.5 1 FIGURE 1.4.21 On the interval [−1, 1] it’s wrong. -0.5 0 t FIGURE 1.4.22 On the interval [−1/2, 1/2] it’s bizarre. 0.5 -0.03-0.02-0.01 0 0.01 0.02 0.03 t FIGURE 1.4.23 On the interval [−1/30, 1/30] it’s correct! computer program was used to plot the alleged graphs of I (t) shown in Figs. 1.4.21 through 1.4.23. The graph in Fig. 1.4.21 is plotted on the interval −1 t 1, where we should see 120 complete oscillations because the period of I (t) in Eq. (15) is 1/60 s. But instead the figure shows exactly one oscillation, so something has gone badly wrong. The graph in Fig. 1.4.22 is plotted on the interval − 12 t 12 , and whatever it is has gone from merely wrong to outright bizarre. Finally, in Fig. 1.4.23 1 1 4 t 30 of length 60 , so we should see exactly the graph is plotted on the interval − 30 4 complete oscillations. And indeed we do, so we’ve finally got a correct graph of the ◗ current function in Eq. (15). Here’s an explanation of what went wrong at first in Example 13. The computer was programmed to plot values at exactly 120 equally spaced points of the interval desired. So in Fig. 1.4.21 we’re plotting only 1 point per cycle—not nearly enough to capture the actual shape of the curve—and only 2 points per cycle in Fig. 1.4.22. But in Fig. 1.4.23 we’re plotting 30 points per cycle, and this gives an accurate representation of the actual graph. The incorrect graph in Fig. 1.4.21—which seems to portray an oscillation with 1 s—is an example of the phethe incorrect period of 2 s, instead of the correct 60 nomenon of aliasing. Another example of aliasing, occasionally seen in old Western movies, is the wagon wheel that appears to rotate slowly in the wrong direction. 1 REMARK The aliasing phenomenon exhibited in Figs. 1.4.21 and 1.4.22 is heavily dependent on the precise number of points being plotted. A plotting device (such as graphing calculator) that uses a fixed number of plotting points is susceptible to aliasing. More sophisticated graphing utilities may avoid aliasing by using a variable number of nonuniformly spaced plotting points. 0.5 I 0 -0.5 -1 -0.5 0 t 0.5 FIGURE 1.4.24 Individual plotted points that are joined by line segments in Fig. 1.4.22. Figure 1.4.22 consists largely of line segments joining consecutive points that are far apart. Figure 1.4.24 shows how that incorrect graph came about; points 1, 3, 5, 7, . . . , 117, 119 in the interval [−0.5, 0.5] are plotted in red, whereas points 2, 4, 6, . . . , 118, 120 are plotted in blue. Now you can see what happened when the computer plotted line segments joining point 1 to point 2, point 2 to point 3, and so forth. One moral of Example 13 is that it pays to know what you’re looking for in a graph. If the graph looks markedly different in windows of different sizes, this is a clue that something’s wrong. Whereas in Example 13 we got anomalous results by plotting the graph in windows of different sizes, the next example illustrates a situation where we must plot graphs on different scales in order to see the whole picture. EXAMPLE 14 Now suppose that a high-frequency (6000 Hz) current of 0.01 ampere is added to the current in Eq. (15), so the resulting current is described by I (t) = sin 120π t + (0.01) sin 12000πt. 42 (16) Transcendental Functions SECTION 1.4 43 0.2 0.15 1 0.1 0.5 I 0.05 I 0 0 -0.05 -0.5 -0.1 -1 -0.15 -0.01 0 t 0.01 FIGURE 1.4.25 I (t) = sin 120π t + 0.01 sin 12000πt on the interval −1/60 t 1/60. -0.2 -4 -3 -2 -1 0 t 1 2 3 4 × 10 -4 FIGURE 1.4.26 I (t) = sin 120π t + 0.01 sin 12000π t on the interval −1/2400 t 1/2400. 1 1 When we plot Eq. (16) on the interval − 60 t 60 , we get the graph shown in Fig. 1.4.25. It looks like two cycles of the original current in (15), although the plot is perhaps a bit “fuzzy.” To see the effect of the added second term in Eq. (16) we must plot the graph on a much magnified scale, as in Fig. 1.4.26. The “fuzz” in Fig. 1.4.25 has now been magnified to show clearly the high-frequency oscillations with period 1 s. ◗ 6000 1.4 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. For every real number x, sin(x + 2π ) = sin x. 2. The equation cos x = 0 has no solutions. 3. The composition h = f ◦ g of the functions f and g has the formula h(x) = f (g(x)). 4. If f and g are functions, then f ◦ g = g ◦ f . 5. If f (x) = x 2 and g(x) = cos x, then f (g(x)) = cos(x 2 ). 6. If f (x) = 2x , then f (x) → −∞ as x → −∞. 7. The statement y = loga x means that a y = x. 8. To the number of digits shown, e ≈ 2.71828. 9. The equation x = cos x has infinitely many real solutions. 10. If x > 100,000, then log x < 12 x 1/5 . 1.4 CONCEPTS: QUESTIONS AND DISCUSSION Each of the following items describes a particular population numbering P(t) at time t. Tell whether you think the function P(t) seems more likely to be a linear, quadratic, polynomial, root, rational, trigonometric, exponential, or logarithmic function of t. In each case write a specific function satisfying the given description. 1. The population triples every five years. 2. The population increases by the same amount each year. 3. The population oscillates every five years between a maximum of 120 and a minimum of 80. 4. The population decreases for a time, reaches a minimum value, then increases thereafter (getting larger and larger as time goes on). 5. The population increases for a time and reaches a maximum value, then decreases for a time and reaches a minimum value, and thereafter increases (becoming larger and larger). 43 44 CHAPTER 1 Functions, Graphs, and Models 6. The population increases each year, but by a smaller percentage than it increased in the preceding year. 7. The population decreases by the same percentage each year. 8. The population increases for a time, reaches a maximum value, and decreases thereafter (apparently dying out) with P(t) approaching zero as t increases. 1.4 PROBLEMS In Problems 1 through 10, match the given function with its graph among those shown in Figs. 1.4.27 through 1.4.36. Try to do this without using your graphing calculator or computer. 1. f (x) = 2x − 1 2. f (x) = 2 − 3−x 3. f (x) = 1 + cos x 4. f (x) = 2 − 2 sin x 5. f (x) = 1 + 2 cos x 6. f (x) = 2 − sin x 7. f (x) = x 2x 9. f (x) = 1 + cos 6x 1 + x2 8. f (x) = 3 4 1.5 y 1 y 1 0 log x x -2 -2 0 2 4 x 6 8 -1 -6 10 1 -1 0 -2 FIGURE 1.4.27 0 x 2 4 6 1 0.5 2 0 10 -2 FIGURE 1.4.34 3 y 2 5 -4 4 y 1 y 1 y 0 0 -1 -10 -3 -10 -5 0 x 5 -0.5 -5 10 0 x 5 -1 10 FIGURE 1.4.35 FIGURE 1.4.28 0 5 10 15 20 25 x FIGURE 1.4.36 In Problems 11 through 20, find f (g(x)) and g( f (x)). 4 11. f (x) = 1 − x 2 , g(x) = 2x + 3 3 1 2 0.5 y 1 y 0 0 -0.5 -1 -1 -2 -3 -2 -1 0 x 1 2 3 12. f (x) = −17, g(x) = |x| √ 13. f (x) = x 2 − 3, g(x) = x 2 + 3 1 14. f (x) = x 2 + 1, g(x) = 2 x +1 √ 15. f (x) = x 3 − 4, g(x) = 3 x + 4 √ 16. f (x) = x, g(x) = cos x 0 1 2 3 4 5 x FIGURE 1.4.29 19. f (x) = 1 + x 2 , g(x) = tan x 6 20. f (x) = 1 − x 2 , g(x) = sin x 5 In Problems 21 through 30, find a function of the form f (x) = x k (you must specify k) and a function g such that f (g(x)) = h(x). 4 0.5 3 y 2 y 1 0 0 -1 0 2 4 6 8 10 -2 -10 -5 x FIGURE 1.4.31 17. f (x) = sin x, g(x) = x 3 18. f (x) = sin x, g(x) = cos x FIGURE 1.4.30 1 -0.5 -0.5 FIGURE 1.4.33 3 0 x 0.5 -1 2 -1 44 2 2 5 5 -5 2.5 3 0 10. f (x) = 2−x sin 10x 4 -2 -10 3 4 FIGURE 1.4.32 0 x 5 10 21. h(x) = (2 + 3x)2 √ 23. h(x) = 2x − x 2 22. h(x) = (4 − x)3 25. h(x) = (5 − x 2 )3/2 1 27. h(x) = x +1 1 29. h(x) = √ x + 10 26. h(x) = 24. h(x) = (1 + x 4 )17 (4x − 6)4 1 28. h(x) = 1 + x2 1 30. h(x) = (1 + x + x 2 )3 3 Preview: What Is Calculus? SECTION 1.5 45 In Problems 31–40, use a graphing calculator or computer to determine the number of real solutions by inspecting the graph of the given equation. 31. x = 2−x 32. x + 1 = 3 cos x 33. x − 1 = 3 cos x 34. x = 5 cos x 35. x = 7 cos x 36. 2 log10 x = cos x 37. log10 x = cos x (x > 0) (x > 0) 38. x 2 = 10 cos x 39. x = 100 sin x 2 40. x = 5 cos x + 10 log10 x (x > 0) 41. Consider the population of Example 7 in this section, which starts with 10 rodents and doubles every month. Determine graphically (that is, by zooming) how long it will take this population to grow to 100 rodents. (Assume that each month is 30 days long and obtain an answer correct to the nearest day.) 42. Consider the money-market account of Example 8, which pays 8% annually. Determine graphically how long it will take the initial investment of $5000 to triple. 43. In 1980 the population P of Mexico was 67.4 million and was growing at the rate of 2.6% per year. If the population continues to grow at this rate, then t years after 1980 it will be P(t) = 67.4 · (1.026)t (millions). Determine graphically how long it will take the population of Mexico to double. about 2.5% of the strontium-90 disappears each year. Then the amount of radiation left after t years will be A(t) = 12 · (0.975)t (measured in “safe units” of radiation). Determine graphically how long (after the original accident) it will be until the region measures only 1 safe unit, and it is therefore safe for humans to return. 46. Refer to Example 6 of this section; determine graphically the value (accurate to three decimal places) of the negative solution of the equation x 2 = 2x . 47. Refer to Example 10 of this section; determine graphically the value (accurate to three decimal places) of the solution near x = 5 of the equation log10 x = 12 x 1/5 . 48. The equation x 10 = 3x has three real solutions. Graphically approximate each of them accurate to two decimal places. 49. You land your space ship on a spherical asteroid between Earth and Mars. Your copilot walks 1000 feet away along the asteroid’s smooth surface carrying a 10-ft rod and thereby vanishes over the horizon. When she places one end of the rod on the ground and holds it straight up and down, you— lying flat on the ground—can just barely see the tip of the rod. Use this information to find the radius R of this asteroid (in miles). The key will be to derive a pair of equations relating R and the angle θ indicated in Fig. 1.4.37. (Think of the right triangle shown there and of the relationship between circular arc length and subtended central angle.) You can then attempt to solve these equations graphically. You should find plenty of solutions. But which of them gives the radius of the asteroid? 44. Suppose that the amount A of ozone in the atmosphere decreases at the rate of 0.25% per year, so that after t years the amount remaining is A(t) = A 0 (0.9975)t , where A 0 denotes the initial amount. Determine graphically how long it will take for only half the original amount of ozone to be left. Does the numerical value of A 0 affect this answer? 45. The nuclear accident at Chernobyl left the surrounding region contaminated with strontium-90, which initially was emitting radiation at approximately 12 times the level safe for human habitation. When an atom of strontium-90 emits radiation, it decays to a nonradioactive isotope. In this way, 10 ft 1000 ft θ R FIGURE 1.4.37 The asteroid problem. 1.5 PREVIEW: WHAT IS CALCULUS? Surely this question is on your mind as you begin a study of calculus that may extend over two or three terms. Following our review of functions and graphs in Sections 1.1 through 1.4, we can preview here at least the next several chapters, where the central concepts of calculus are developed. The Two Fundamental Problems The body of computational technique that constitutes “the calculus” revolves around two fundamental geometric problems that people have been investigating for more than 2000 years. Each problem involves the graph y = f (x) of a given function. The first fundamental problem is this: What do we mean by the line tangent to the curve y = f (x) at a given point? The word tangent stems from the Latin word tangens, for “touching.” Thus a line tangent to a curve is one that “just touches” the curve. Lines tangent to circles (Fig. 1.5.1) are well known from elementary geometry. 45 46 CHAPTER 1 Functions, Graphs, and Models 4 y = x2 2 y P (1, 1) 0 y = 2x − 1 L -2 -4 FIGURE 1.5.1 The tangent line L touches the circle at the point P. y y = f(x) L P(x, f (x)) -2 0 x 2 4 FIGURE 1.5.2 The line tangent to the parabola y = x 2 at the point (1, 1). Figure 1.5.2 shows the line tangent to the parabola y = x 2 at the point (1, 1). We will see in Section 2.1 that this particular tangent line has slope 2, so its point-slope equation is y − 1 = 2 · (x − 1); that is, y = 2x − 1. Our first problem is how to find tangent lines in more general cases. x FIGURE 1.5.3 What is the slope of the line L tangent to the graph y = f (x) at the point P(x, f (x))? The Tangent Problem Given a point P(x, f (x)) on the curve y = f (x), how do we calculate the slope of the tangent line at P (Fig. 1.5.3)? We begin to explore the answer to this question in Chapter 2. If we denote by m(x) the slope of the tangent line at P(x, f (x)), then m is a new function. It might informally be called a slope-predictor for the curve y = f (x). In calculus this slopepredictor function is called the derivative of the function f . In Chapter 3 we learn to calculate derivatives of a variety of functions, and in both Chapter 3 and Chapter 4 we see numerous applications of derivatives in solving real-world problems. These three chapters introduce part of calculus called differential calculus. The tangent problem is a geometric problem—a purely mathematical question. But its answer (in the form of derivatives) is the key to the solution of diverse applied problems in many scientific and technical areas. Examples 1 and 2 may suggest to you the connections that are the key to the pivotal role of calculus in science and technology. EXAMPLE 1 Suppose that you’re driving a car along a long, straight road (Fig. 1.5.4). If f (t) denotes the distance (in miles) the car has traveled at time t (in hours), then the slope of the line tangent to the curve y = f (t) at the point (t, f (t)) ◗ (Fig. 1.5.5) is the velocity (in miles per hour) of the car at time t. y Distance y = f(t) Distance f(t) Start Time t FIGURE 1.5.4 A car on a straight road (Example 1). 46 (t, f (t)) Time t FIGURE 1.5.5 The slope of the tangent line at the point (t, f (t)) is the velocity at the time t (Example 1). Preview: What Is Calculus? SECTION 1.5 47 EXAMPLE 2 Suppose that f (t) denotes the number of people in the United States who have a certain serious disease at time t (measured in days from the beginning of the year). Then the slope of the line tangent to the curve y = f (t) at the point (t, f (t)) (Fig. 1.5.6) is the rate of growth (the number of persons newly affected per day) of the ◗ diseased population at time t. y Population y = f(t) (t, f(t)) Time t FIGURE 1.5.6 The rate of growth of f (t) at the time t is the slope of the tangent line at the point (t, f (t)) (Example 2). NOTE The truth of the statements made in these two examples is not obvious. To understand such things is one reason you study calculus! We return to the concepts of velocity and rate of change at the beginning of Chapter 3. Here we will be content with the observation that the slopes of the tangent lines in Examples 1 and 2 at least have the correct units. If in the time-distance plane of Example 1 we measure time t (on the horizontal axis) in seconds and distance y (on the vertical axis) in feet (or meters), then the slope (ratio of rise to run) of a straight line has the dimensions of feet (or meters) per second—the proper units for velocity (Fig. 1.5.7). Similarly, if in the t y-plane of Example 2 time t is measured in months and y is measured in persons, then the slope of a straight line has the proper units of persons per month for measuring the rate of growth of the afflicted population (Fig. 1.5.8). y (persons) y (ft) Slope units: ft s Slope units: persons month Rise (persons) Rise (ft) Run (months) Run (s) t (months) t (s) FIGURE 1.5.7 Here slope has the dimensions of velocity (ft/s). FIGURE 1.5.8 Here slope has the dimensions of rate of change of population. The second fundamental problem of calculus is the problem of area. Given the graph y = f (x), what is the area between this graph and the x-axis over the interval [a, b]? The Area Problem If f (x) 0 for x in the interval [a, b], how do we calculate the area A of the plane region that lies between the curve y = f (x) and the x-axis over the interval [a, b] (Fig. 1.5.9)? y y = f (x) Area A = ? a b x FIGURE 1.5.9 The area problem. We begin to explore the answer to this second question in Chapter 5. In calculus the area A is called an integral of the function f . Chapters 5 and 6 are devoted to the calculation and application of integrals. These two chapters introduce the other part of calculus, which is called integral calculus. Like the tangent problem, the area problem is a purely mathematical question, but its answer (in the form of integrals) has extensive ramifications of practical importance. Examples 3 and 4 have an obvious kinship with Examples 1 and 2. 47 48 CHAPTER 1 Functions, Graphs, and Models EXAMPLE 3 If f (t) denotes the velocity of a car at time t, then the area under the curve y = f (t) over the time interval [a, b] is equal to the distance traveled by the car ◗ between time t = a and time t = b (Fig. 1.5.10). y Velocity y = f (t) Area A a b Time t FIGURE 1.5.10 The area A under the velocity curve is equal to the distance traveled during the time interval a t b (Example 3). EXAMPLE 4 If f (t) denotes the rate of growth of a diseased population at time t, then the area under the curve y = f (t) over the time interval [a, b] is equal to the net change in the size of this population between time t = a and time t = b (Fig. 1.5.11). ◗ When we discuss integrals in Chapter 5, you will learn why the statements in Examples 3 and 4 are true. The Fundamental Relationship y Rate of change y = f (t) Area A a b Time t FIGURE 1.5.11 The area A under the rate-of-change curve is equal to the net change in the population from time t = a to t = b (Example 4). Examples 1 and 3 are two sides of a certain coin: There is an “inverse relationship” between the distance traveled and the velocity of a moving car. Examples 2 and 4 exhibit a similar relationship between the size of a population and its rate of change. Both the distance/velocity relationship and the size/rate-of-change relationship illustrated by Examples 1 through 4 are consequences of a deep and fundamental relationship between the tangent problem and the area problem. This more general relationship is described by the fundamental theorem of calculus, which we discuss in Chapter 5. It was discovered in 1666 by Isaac Newton at the age of 23 while he was still a student at Cambridge University. A few years later it was discovered independently by Gottfried Wilhelm Leibniz, who was then a German diplomat in Paris who studied mathematics privately. Although the tangent problem and the area problem had, even then, been around for almost 2000 years, and much progress on separate solutions had been made by predecessors of Newton and Leibniz, their joint discovery of the fundamental relationship between the area and tangent problems made them famous as “the inventors of the calculus.” Applications of Calculus So calculus centers around the computation and application of derivatives and integrals—that is, of tangent line slopes and areas under graphs. Throughout this textbook, you will see concrete applications of calculus to different areas of science and technology. The following list of a dozen such applications gives just a brief indication of the extraordinary range and real-world power of calculus. • • • • • • • 48 Suppose that you make and sell tents. How can you make the biggest tent from a given amount of cloth and thereby maximize your profit? (Section 3.6) You throw into a lake a cork ball that has one-fourth the density of water. How deep will it sink in the water? (Section 3.10) A driver involved in an accident claims he was going only 25 mi/h. Can you determine from his skid marks the actual speed of his car at the time of the accident? (Section 5.2) The great pyramid of Khufu at Gizeh, Egypt, was built well over 4000 years ago. No personnel records from the construction remain, but nevertheless we can calculate the approximate number of laborers involved. (Section 6.5) Suppose that you win the Florida lottery and decide to use part of your winnings to purchase a “perpetual annuity” that will pay you and your heirs (and theirs, ad infinitum) $10,000 per year. What is a fair price for an insurance company to charge you for such an annuity? (Section 7.8) If the earth’s population continues to grow at its present rate, when will there be “standing room only”? (Section 8.1) The factories polluting Lake Erie are forced to cease dumping wastes into the lake immediately. How long will it take for natural processes to restore the lake to an acceptable level of purity? (Section 8.4) Preview: What Is Calculus? SECTION 1.5 49 • • • • • • In 1845 the Belgian demographer Verhulst used calculus to predict accurately the course of U.S. population growth (to within 1%) well into the twentieth century, long after his death. How? (Section 8.5) What explains the fact that a well-positioned reporter can eavesdrop on a quiet conversation between two diplomats 50 feet away in the Whispering Gallery of the U.S. Senate, even if this conversation is inaudible to others in the same room? (Section 9.6) Suppose that Paul and Mary alternately toss a fair six-sided die in turn until one wins the pot by getting the first “six.” How advantageous is it to be the one who tosses first? (Section 10.3) How can a submarine traveling in darkness beneath the polar icecap keep accurate track of its position without being in radio contact with the rest of the world? (Section 11.5) Suppose that your club is designing an unpowered race car for the annual downhill derby. You have a choice of solid wheels, bicycle wheels with thin spokes, or even solid spherical wheels (like giant ball bearings). Can you determine (without time-consuming experimentation) which will make the race car go the fastest? (Section 13.5) Some bullets have flattened tips. Is it possible that an artillery shell with a flat-tipped “nose cone” may experience less air resistance—and therefore travel farther—than a shell with a smoothly rounded tip? (Section 14.5) 1.5 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The tangent problem is the problem of finding the slope of the straight line tangent to the graph of y = f (x) at the point P of the graph. 2. The area problem is the problem of finding the area of the plane region above the x-axis and below the graph of the function y = f (x) 0 for a x b. 3. The fundamental theorem of calculus was discovered by Newton and, independently, by Leibniz. 4. The slope of the line tangent to the graph of y = x 2 at the point (1, 1) is 4. 5. If a straight line touches or intersects a curve at exactly one point, then it is tangent to the curve at that point. 6. If a straight line touches or intersects a curve at more than one point, then it cannot be tangent to the curve at any of those points. 7. A function that predicts the slope of the line tangent to the graph of the function f at the point (x, f (x)) is called the derivative of f . 8. The computation of area is one topic studied in integral calculus. 9. The relation between distance and velocity has nothing to do with calculus. 10. The fundamental theorem of calculus shows that the tangent problem and the area problem are related. 49 50 CHAPTER 1 Functions, Graphs, and Models CHAPTER 1: REVIEW Understanding: Concepts and Definitions Refer to the listed pages to review the concepts and definitions in this chapter that you need to understand. Section Pages 1.1 The definition of a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 The domain and range of a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Dependent and independent variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Open and closed interval notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 What is a formula vs. what is a relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 The idea of a mathematical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2 Slope-intercept and point-slope equations of straight lines . . . . . . . . . . . . . . . . . . . . . . . . . . 12 The graph of an equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Circles and translates of graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13–14 The graph of a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 The vertical line test for graphs of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Discontinuities of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Parabolas and graphs of quadratic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17–18 Graphic, numeric, and symbolic representations of functions . . . . . . . . . . . . . . . . . . . . . . . . 20 1.3 The definition of a power function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Algebraic combinations of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 The definition of a polynomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 The definition of a rational function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 The definition of an algebraic function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 1.4 The sine and cosine functions and their graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 The definition of the composition of two functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 The definition of an exponential function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38 The definition of a logarithmic function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Objectives: Methods and Techniques Work the listed problems in each section to practice the methods and techniques in this chapter that you need to master. Section Problems 1.1 Simplifying functional expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13, 15 Finding the domain of a function defined by a formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25, 29, 33 Writing formulas for functions described verbally . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37, 39, 41, 43, 45 Numerical solution of equations by repeated tabulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59, 60 1.2 Writing the equation of a given straight line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 5, 9 Sketching the graph of a circle with given equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13, 15 Sketching a parabola with given equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Identifying and sketching the graph of a given function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33, 37, 39, 45, 49 Algebraic and graphical investigation of high and low points . . . . . . . . . . . . . . . . . . . . . . . . 57, 61 1.3 Finding formulas for algebraic combinations of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 5 Identifying the graph of a polynomial by determining its number of zeros . . . . . . . . . . . . 7, 11 and its behavior for |x| large Identifying the graph of a rational function by determining its asymptotes . . . . . . . . . . . . 13, 15 and its behavior for |x| large Finding graphically the number of real zeros of a polynomial . . . . . . . . . . . . . . . . . . . . . . . 21, 23, 25, 39 1.4 Matching graphs and equations of trigonometric and exponential functions . . . . . . . . . . . 1, 3, 5, 7 Finding the formula for the composition f (g(x)) of two given functions f and g . . . . . 11, 15, 17, 19 Finding graphically the number of real solutions of a given transcendental equation . . . 31, 33, 35, 39 50 Chapter 1 Miscellaneous Problems 51 MISCELLANEOUS PROBLEMS In Problems 1 through 10, find the domain of definition of the function with the given formula. 3. 5. 7. 9. 11. 12. √ 1 2−x x 1 4. f (x) = 2 f (x) = 2 x −9 x +1 √ 3 x +1 f (x) = 1 + x 6. f (x) = 2 x − 2x √ 1 f (x) = 2 − 3x 8. f (x) = √ 9 − x2 √ f (x) = (x − 2)(4 − x) 10. f (x) = (x − 2)(4 − x) In accord with Boyle’s law, the pressure p (lb/in.2 ) and volume V (in.3 ) of a certain gas satisfy the condition pV = 800. What is the range of possible values of p, given 100 V 200? The relationship between the Fahrenheit temperature F and the Celsius temperature C is given by 1. f (x) = x −4 2. f (x) = F = 32 + 95 C. If the temperature on a given day ranges from a low of 70◦ F to a high of 90◦ F, what is the range of temperature in degrees Celsius? 13. An electric circuit contains a battery that supplies E volts in series with a resistance of R ohms (Fig. 1.MP.1). Then the current of I amperes that flows in the circuit satisfies Ohm’s law, E = I R. If E = 100 and 25 < R < 50, what is the range of possible values of I ? Current: I amperes Battery: E volts Resistance: R ohms 20. L passes through (4, −1) and has slope −3. 1 2 21. L has slope and y-intercept −5. 22. L passes through (2, −3) and is parallel to the line with equation 3x − 2y = 4. 23. L passes through (−3, 7) and is perpendicular to the line with equation y − 2x = 10. (Appendix B reviews slopes of perpendicular lines.) 24. L is the perpendicular bisector of the segment joining (1, −5) and (3, −1). In Problems 25 through 34, match the given function with its graph among those shown in Figs. 1.MP.2 through 1.MP.11. Do this without using your graphing calculator or computer. Instead, rely on your knowledge of the general characteristics of polynomial, rational, algebraic, trigonometric, exponential, and logarithmic functions. 2 1.5 1 0.5 y 0 -0.5 -1 -1.5 -2 -2 2 1.5 1 y 0.5 0 -0.5 0 2 4 x 6 8 -1 -10 10 FIGURE 1.MP.2 In Problems 19 through 24, write an equation of the straight line L described. 19. L passes through (−3, 5) and (1, 13). 10 5 10 10 5 20 y 0 y 0 -20 -5 -40 -10 -60 -5 0 x -15 -10 5 FIGURE 1.MP.4 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -5 5 15 40 14. The period T (in seconds) of √ a simple pendulum of length L (in feet) is given by T = 2π L/32. If 3 < L < 4, what is the range of possible values of T ? 15. Express the volume V of a cube as a function of its total surface area S. 16. The height of a certain right circular cylinder is equal to its radius. Express its total surface area A (including both ends) as a function of its volume V . 17. Express the area A of an equilateral triangle as a function of its perimeter P. 18. A piece of wire 100 in. long is cut into two pieces of lengths x and 100 − x. The first piece is bent into the shape of a square, the second into the shape of a circle. Express as a function of x the sum A of the areas of the square and circle. 0 x FIGURE 1.MP.3 60 FIGURE 1.MP.1 The simple electric circuit of Problem 13. -5 -5 0 x FIGURE 1.MP.5 10 5 y 0 0 x FIGURE 1.MP.6 5 -5 -5 0 x 5 FIGURE 1.MP.7 51 52 CHAPTER 1 Functions, Graphs, and Models 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -10 -5 0 x 5 10 10 8 6 4 2 y 0 -2 -4 -6 -8 -10 -5 FIGURE 1.MP.8 FIGURE 1.MP.9 5 4 3 2 1 y 0 -1 -2 -3 -4 -5 -5 10 8 6 4 2 y 0 -2 -4 -6 -8 -10 -5 0 x 5 FIGURE 1.MP.10 the two cases separately to conclude that the solution set is (−∞, −2) ∪ (3, +∞). ] Use the method of Problem 47 to solve the inequalities in Problems 48 through 50. 48. x 2 − 3x + 2 < 0 50. 2x 15 − x 2 0 x 5 27. f (x) = x 4 − 4x 3 + 5 5 29. f (x) = 2 x −x +6 31. f (x) = 2−x − 1 33. f (x) = 1 + 3 sin x 0 x 5 26. f (x) = x 3 − 4x 2 + 5 5 28. f (x) = 2 x −x −6 √ 30. f (x) = 8 + 2x − x 2 32. f (x) = log10 (x + 1) 34. f (x) = x + 3 sin x Sketch the graphs of the equations and functions given in Problems 35 through 44. 35. 2x − 5y = 7 36. |x − y| = 1 2 2 37. x + y = 2x 38. x 2 + y 2 = 4y − 6x + 3 2 40. y = 4x − x 2 39. y = 2x − 4x − 1 1 1 42. f (x) = 41. f (x) = x +5 4 − x2 43. f (x) = |x − 3| 44. f (x) = |x − 3| + |x + 2| 45. Apply the triangle inequality (of Appendix A) twice to show that |a + b + c| |a| + |b| + |c| for arbitrary real numbers a, b, and c. 46. Write a = (a − b) + b to deduce from the triangle inequality (of Appendix A) that |a| − |b| |a − b| for arbitrary real numbers a and b. 47. Solve the inequality x 2 − x −6 > 0. [ Suggestion: Conclude from the factorization 51. x 2 − 5x − 7 = 0 52. 3x 2 − 10x − 11 = 0 53. 4x − 14x + 11 = 0 54. 5x 2 + 24x − 35 = 0 55. 8x 2 + 33x − 36 = 0 56. 9x 2 + 74x − 156 = 0 In Problems 57 through 62, apply either the method of repeated tabulation or the method of successive zooms (or both) to find the lowest point on the given parabola. You may check your work by completing the square. 57. y = x 2 − 5x + 7 58. y = 3x 2 − 10x + 11 59. y = 4x − 14x + 11 60. y = 5x 2 + 24x + 35 61. y = 8x 2 + 33x + 35 62. y = 9x 2 + 74x + 156 2 63. Figure 1.MP.12 shows a 10-cm by 7-cm portrait that includes a border of width x on the top and bottom and of width 2x on either side. The area of the border is itself 20 cm2 . Use either repeated tabulation or successive zooms to find x. x 2x x 10 cm FIGURE 1.MP.12 The bordered portrait of Problem 63. 64. A mail-order catalog lists a 60-in. by 35-in. tablecloth that shrinks 7% in area when first washed. The catalog description also implies that the length and width will both decrease by the same amount x. Use numerical (tabulation) or graphical (zoom) methods to find x. Determine graphically the number of real solutions of each equation in Problems 65 through 70. 65. x 3 − 7x + 3 = 0 66. x 4 − 3x 2 + 4x − 5 = 0 67. sin x = x 3 − 3x + 1 68. cos x = x 4 − x that the quantities x − 3 and x + 2 must be either both positive or both negative for the inequality to hold. Consider 69. cos x = log10 x PHOTO CREDITS 1 Corbis/Bettmann 2x 7 cm x − x − 6 = (x − 3)(x + 2) 2 52 The remaining problems require the use of an appropriate calculator or computer. In Problems 51 through 56, use either the method of repeated tabulation or the method of successive zooms (or both) to find the two roots (with three digits to the right of the decimal point correct or correctly rounded) of the given quadratic equation. You may check your work with the aid of the quadratic formula and an ordinary calculator. 2 FIGURE 1.MP.11 25. f (x) = 2 − 2x − x 2 49. x 2 − 2x − 8 > 0 70. 10−x = log10 x 2 Prelude to Calculus T he modern computer programming language Ada is named in honor of Ada Byron, daughter of the English poet Lord Byron. Her interest in science and mathematics led her around 1840 to study the Difference Engine, a gear-based mechanical calculator that the mathematician Charles Babbage Ada Byron (1815–1852) had built to compute tables of values of functions. By then he was designing his much more advanced Analytic Engine, an elaborate computing machine that would have been far ahead of its time if it had been completed. In 1843 Ada Byron wrote a series of brief essays explaining the planned operation of the Analytical Engine and its underlying mathematical principles. She included a prototype “computer program” to illustrate how its calculations were to be “programmed” in advance, using a deck of punched cards to specify its instructions. Calculus has been called “the calculating engine par excellence.” But in our own time the study and applications of calculus have been reshaped by electronic computers. Throughout this book we illustrate concepts of calculus by means of graphic, numeric, and symbolic results generated by computers. In Chapter 2 we exploit computational technology systematically in the investigation of limits. Grace Murray Hopper (1906–1992) The Difference Engine Almost exactly a century after the death of Ada Byron, the first modern computer compiler (for translation of human-language programs into machine-language instructions) was developed by Grace Murray Hopper. As a mathematician and U.S. Navy officer, Hopper had worked with the very first modern electronic computers developed during and immediately after World War II. In 1967 she was recalled to active duty to lead efforts to standardize the computer language COBOL for the Navy. In 1985 at the age of 79, she became Rear Admiral Grace Hopper. In 1986 she was retired—as the Navy’s oldest commissioned officer on active duty—in a ceremony held aboard the U.S.S. Constitution, the Navy’s oldest commissioned warship. From Chapter 2 of Calculus, Early Transcendentals, Seventh Edition. C. Henry Edwards, David E. Penney. Copyright © 2008 by Pearson Education, Inc. All rights reserved. 53 54 CHAPTER 2 Prelude to Calculus 2.1 TANGENT LINES AND SLOPE PREDICTORS P O FIGURE 2.1.1 The line tangent to the circle at the point P is perpendicular to the radius OP. In Sections 1.2 and 1.5 we saw that certain applied problems raise the question of what is meant by the tangent line at a specified point of a general curve y = f (x). In this section we see that this “tangent-line problem” leads to the limit concept, which we pursue further in Section 2.2. In elementary geometry the line tangent to a circle at a point P is defined as the straight line through P that is perpendicular to the radius (OP) to that point (Fig. 2.1.1). A general graph y = f (x) has no radius for us to use, but the line tangent to the graph at the point P should be the straight line through P that has—in some sense—the same direction at P as the curve itself. Because a line’s “direction” is determined by its slope, our plan for defining a line tangent to a curve amounts to finding an appropriate “slope-prediction formula” that will give the proper slope of the tangent line. Example 1 illustrates this approach in the case of one of the simplest of all nonstraight curves, the parabola with equation y = x 2 . EXAMPLE 1 Determine the slope of the line L tangent to the parabola y = x 2 at the point P(a, a 2 ). Solution Figure 2.1.2 shows the parabola y = x 2 and a typical point P(a, a 2 ) on it. The figure also shows a visual guess of the direction of the desired tangent line L at P. We must find the slope of L. y y = x2 L P(a, a2 ) x FIGURE 2.1.2 The tangent line at P should have the same direction as the curve does at P (Example 1). We cannot immediately calculate the slope of L, because we know the coordinates of only one point P(a, a 2 ) of L. Hence we begin with another line whose slope we can compute. Figure 2.1.3 shows the secant line K that passes through the point P and the nearby point Q(b, b2 ) of the parabola y = x 2 . Let us write h = x = b − a for the difference of the x-coordinates of P and Q. (The notation x is as old as calculus itself, and it means now what it did 300 years ago: an increment, or change, in the value of x.) Then the coordinates of Q are given by the formulas b =a+h and b2 = (a + h)2 . Hence the difference in the y-coordinates of P and Q is y = b2 − a 2 = (a + h)2 − a 2 . Because P and Q are two different points, we can use the definition of slope to calculate the slope m P Q of the secant line K through P and Q. If you change the value of 54 Tangent Lines and Slope Predictors SECTION 2.1 55 y y = x2 K Q(b, b 2 ) L Δy P(a, a2 ) Δx b=a+h a x FIGURE 2.1.3 The secant line K passes through the two points P and Q, which we can use to determine its slope (Example 1). h = x, you change the line K and thereby change its slope. Therefore, m P Q depends on h: mPQ = = (a + h)2 − a 2 y = x (a + h) − a (a 2 + 2ah + h 2 ) − a 2 2ah + h 2 h(2a + h) = = . h h h (1) Because h is nonzero, we may cancel it in the final fraction. Thus we find that the slope of the secant line K is given by m P Q = 2a + h. (2) Now imagine what happens as you move the point Q along the curve closer and closer to the point P. (This situation corresponds to h approaching zero.) The line K still passes through P and Q, but it pivots around the fixed point P. As h approaches zero, the secant line K comes closer to coinciding with the tangent line L. This phenomenon is suggested in Fig. 2.1.4, which shows the secant line K approaching the tangent line L. Our idea is to define the tangent line L as the limiting position of the secant line K . To see precisely what this means, examine what happens to the slope of K as K y y = x2 K L Q P a x FIGURE 2.1.4 As h → 0, Q approaches P, and K moves into coincidence with the tangent line L (Example 1). 55 56 CHAPTER 2 Prelude to Calculus pivots into coincidence with L: As h approaches zero, Q approaches P, and so K approaches L; meanwhile, the slope of K approaches the slope of L. Hence our question is this: As the number h approaches zero, what value does the slope m P Q = 2a + h approach? We can state this question of the “limiting value” of 2a + h by writing lim (2a + h) = ? (3) h→0 Here, “lim” is an abbreviation for the word “limit,” and “h → 0” is an abbreviation for the phrase “h approaches zero.” Thus Eq. (3) asks, “What is the limit of 2a + h as h approaches zero?” For any specific value of a we can investigate this question numerically by calculating values of 2a + h with values of h that become closer and closer to zero—such as the values h = 0.1, h = −0.01, h = 0.001, h = −0.0001, . . . , or the values h = 0.5, h = 0.1, h = 0.05, h = 0.01, . . . . For instance, the tables of values in Figs. 2.1.5 and 2.1.6 indicate that with a = 1 and a = −2 we should conclude that lim (2 + h) = 2 h→0 and lim (−4 + h) = −4. h→0 More generally, it seems clear from the table in Fig. 2.1.7 that lim m P Q = lim (2a + h) = 2a. h→0 h 2+h 0.1 0.01 0.001 0.0001 ↓ 0 2.1 2.01 2.001 2.0002 ↓ 2 FIGURE 2.1.5 As h → 0 (first column), 2 + h approaches 2 (second column). (4) h→0 h −4 + h 0.5 0.1 0.05 0.01 0.005 0.001 ↓ 0 −3.5 −3.9 −3.95 −3.99 −3.995 −3.999 ↓ −4 FIGURE 2.1.6 As h → 0 (first column), −4 + h approaches −4 (second column). h 2a + h 0.01 0.001 .. . ↓ 0 2a + 0.01 2a + 0.001 .. . ↓ 2a FIGURE 2.1.7 As h → 0 (first column), 2a + h approaches 2a (second column) (Example 1). This, finally, answers our original question: The slope m = m(a) of the line tangent to the parabola y = x 2 at the point (a, a 2 ) is given by m = 2a. (5) ◗ The formula in Eq. (5) is a “slope predictor” for (lines tangent to) the parabola y = x 2 . Once we know the slope of the line tangent to the curve at a given point of the curve, we can then use the point-slope formula to write an equation of this tangent line. EXAMPLE 2 With a = 1, the slope predictor in Eq. (5) gives m = 2 for the slope of the line tangent to y = x 2 at the point (1, 1). Hence an equation of this line is y − 1 = 2(x − 1); 56 that is, y = 2x − 1. Tangent Lines and Slope Predictors SECTION 2.1 57 With a = −3, Eq. (5) gives m = −6 as the slope of the line tangent at (−3, 9), so an equation of the line tangent to the curve at this point is y − 9 = −6(x + 3); y = −6x − 9. that is, ◗ In Fig. 2.1.8 the parabola y = x 2 and its tangent line y = 2x − 1 passing through (1, 1) are both graphed. The relationship between the curve and its tangent line is such that as we “zoom in” on the point of tangency, successive magnifications show less and less of a difference between the curve and the tangent line. This phenomenon is illustrated in Figs. 2.1.9 through 2.1.11. 4 1.4 y = x2 3 y = x2 1.2 2 y y 1 1 0.8 0 0.6 -1 0 2 4 0.6 0.8 x FIGURE 2.1.8 The parabola y = x 2 and its tangent line at P(1, 1). 1 x 1.2 1.4 FIGURE 2.1.9 First magnification. 1.2 1.04 y = x2 1.1 1.02 y 1 y 1 0.98 0.9 0.96 0.8 0.8 0.9 1 x 1.1 1.2 FIGURE 2.1.10 Second magnification. 0.96 0.98 1 x 1.02 1.04 FIGURE 2.1.11 Can you see the difference? REMARK In Example 1 we proceeded as though the concept of a tangent line to a curve were self-evident. The actual meaning of the slope-predictor result m = 2a in Eq. (5) is this: Whatever is meant by the line tangent to the parabola y = x 2 at the point P(a, a 2 ), it can only be the unique straight line through P with slope m = 2a. Thus we must define the line tangent to y = x 2 at P to be the line whose point-slope equation is y − a 2 = 2a(x − a). Pictures like those in Figs. 2.1.8 through 2.1.11 certainly support our conviction that this definition is the correct one. More General Slope Predictors The general case of the line tangent to a curve y = f (x) is scarcely more complicated than the special case y = x 2 of Example 1. Given the function f , suppose that we want to find the slope of the line L tangent to y = f (x) at the point P(a, f (a)). As indicated in Fig. 2.1.12, let K be the secant line passing through the point P and the nearby point Q(a + h, f (a + h)) on the graph. The slope of this secant line is the difference quotient mPQ = f (a + h) − f (a) y = x h (with h = 0). (6) 57 58 CHAPTER 2 Prelude to Calculus y K L y Q Slope: m(a) y = f(x) Q(a + h, f (a + h)) P (a, f (a)) Δ y = f (a + h) − f (a) Δx = h Δy = f (a + h) − f (a) P Δx = h a a+h x y = f (x) FIGURE 2.1.12 As h → 0, Q → P, and the slope of K approaches the slope of the tangent line L. a x a+h FIGURE 2.1.13 The slope of the tangent line at (a, f (a)) is f (a + h) − f (a) m(a) = lim . h→0 h We now force Q to approach the fixed point P along the curve y = f (x) by making h approach zero. We ask whether m P Q approaches some limiting value m as h → 0. If so, we write f (a + h) − f (a) m = lim h→0 h and conclude that this number m is the slope of the line tangent to the graph y = f (x) at the point (a, f (a)). Actually, this slope depends on a and we can indicate this by writing m a = lim h→0 f (a + h) − f (a) . h (7) If we can express the limiting value on the right explicitly in terms of a, then Eq. (7) yields a slope predictor for lines tangent to the curve y = f (x). In this case the line tangent to the curve at the point P(a, f (a)) is defined to be the straight line through P that has slope m a . This tangent line is indicated in Fig. 2.1.13. In Chapter 3 we will acknowledge the fact that the slope m a is somehow “derived” from the function f by calling this number the derivative of the function f at the point a. Indeed, much of Chapter 3 will be devoted to methods of calculating derivatives of various familiar functions. Most of these methods are based on the limit techniques of Sections 2.2 and 2.3, but the case of quadratic functions is sufficiently simple for inclusion here. Recall from Section 1.2 that the graph of any quadratic function is a parabola that opens either upward or downward. THEOREM Parabolas and Tangent Lines Consider the parabola y = f (x) where f (x) = px 2 + q x + r (8) (with p = 0). Then the line tangent to this parabola at the point P(a, f (a)) has slope (9) m a = 2 pa + q. Proof The slope of the secant line given in (6) may be simplified as follows: f (a + h) − f (a) [ p(a + h)2 + q(a + h) + r ] − [ pa 2 + qa + r ] = h h 2 2 2 [ p(a + 2ah + h ) + q(a + h) + r ] − [ pa + qa + r ] 2 pah + ph 2 + qh = = , h h m PQ = 58 Tangent Lines and Slope Predictors SECTION 2.1 59 and therefore m P Q = 2 pa + q + ph. The numbers p, q, and a are fixed, so as h → 0 the product ph approaches zero, much as in our computations in Example 1. Thus m a = lim m P Q = lim (2 pa + q + ph) = 2 pa + q, h→0 h→0 ◆ as claimed in Eq. (9). REMARK 1 Thus the formula m a = 2 pa + q provides a ready slope predictor for lines tangent to the parabola with equation y = px 2 + q x + r. Given the coefficients p, q, r , and the number a, we need only substitute in this slopepredictor formula to obtain the slope m a of the line tangent to the parabola at the point where x = a. We need not repeat the computational steps that were carried out in the derivation of the slope-predictor formula. REMARK 2 If we replace a with x we get the slope-predictor function m(x) = 2 px + q. (10) Here m is a function whose value m(x) at x is the slope of the line tangent to the parabola y = f (x) at the point P(x, f (x)). Perhaps the visual scheme f (x) = ⏐ px 2 ⏐ m(x) = 2 px + qx ⏐ + r ⏐ + + 0 q makes this slope predictor easy for you to remember. y y = x2 EXAMPLE 3 Find an equation of the line tangent to the parabola y = 2x 2 − 3x + 5 at the point where x = −1. Solution Here we have p = 2, q = −3, r = 5, and the y-coordinate of our point is 2 · (−1)2 − 3 · (−1) + 5 = 10. Then Eq. (10) gives the slope predictor (c, c2 ) N m(x) = 2 · 2x + (−3) = 4x − 3, x (3, 0) FIGURE 2.1.14 The normal line N from the point (3, 0) to the point (c, c2 ) on the parabola y = x 2 . so the slope of the line tangent to the parabola at the point (−1, 10) is m(−1) = 4 · (−1) − 3 = −7. The point-slope equation of this tangent line is therefore y − 10 = (−7)(x + 1); that is, y = −7x + 3. ◗ Normal Lines y Normal line, slope: −1/m y = f (x) P(c, f (c)) x Tangent line, slope: m FIGURE 2.1.15 The tangent line and normal line through the point P on a curve. How would you find the point P(c, c2 ) that lies on the parabola y = x 2 and is closest to the point (3, 0)? Intuitively, the line segment N with endpoints (3, 0) and P should be perpendicular, or normal, to the parabola’s tangent line at P (Fig. 2.1.14). But if the slope of the tangent line is m, then—by Theorem 2 in Appendix B—the slope of the normal line is 1 (11) mN = − . m (Theorem 2 tells us that if two perpendicular lines have nonzero slopes m 1 and m 2 , then m 1 m 2 = −1.) More precisely, the normal line at a point P of a curve where the tangent line has slope m is defined to be the line through P with slope m N = −1/m (Fig. 2.1.15). Consequently, the parabolic slope predictor in (9) enables us to write equations of lines normal to parabolas as easily as equations of tangent lines. 59 60 CHAPTER 2 Prelude to Calculus EXAMPLE 4 In Example 3 we found that the line tangent to the parabola y = 2x 2 − 3x + 5 at the point P(−1, 10) has slope −7. Therefore the slope of the line normal to that parabola at P is m N = −1/(−7) = 17 . So the point-slope equation of the normal line is y − 10 = 17 (x + 1); x $5/ft y = 17 x + 71 . 7 ◗ The Animal Pen Problem Completed y $5/ft Now we can apply our newfound knowledge of slope-predictor formulas to wrap up our continuing discussion of the animal pen problem of Section 1.1. In Example 9 there we found that the area A of the pen (see Fig. 2.1.16) is given as a function of its base length x by $5/ft y $1/ft x A(x) = 35 (30x − x 2 ) = − 35 x 2 + 18x Wall FIGURE 2.1.16 The animal pen. y 150 that is, Highest point (15, 135) Horizontal tangent line 100 (12) for 0 x 30. Therefore our problem is to find the maximum value of A(x) for x in the closed interval [0, 30]. Let us accept as intuitively obvious—we will see a proof in Chapter 3—the fact that the maximum value of A(x) occurs at the high point where the line tangent to the parabola y = A(x) is horizontal, as indicated in Fig. 2.1.17. But the function A(x) in Eq. (12) is quadratic with p = − 35 and q = 18 (compare (12) with (8)). Therefore the slope predictor in (10) implies that the slope of the tangent line at an arbitrary point (x, A(x)) of the parabola is given by m = m(x) = 2 px + q = − 65 x + 18. y = A(x) 50 We ask when m = 0 and find that this happens when 10 20 30 FIGURE 2.1.17 The graph of y = A(x), 0 x 30. − 65 x + 18 = 0, x and thus when x = 15. In agreement with the result found by algebraic methods in Section 1.2, we find that the maximum possible area of the pen is A(15) = 35 (30 · 15 − 152 ) = 135 (ft2 ). Numerical Investigation of Slopes Suppose that you are given the function f and a specific numerical value of a. You can then use a calculator to investigate the value m = lim h→0 f (a + h) − f (a) h (13) of the slope of the line tangent to the curve y = f (x) at the point (a, f (a)). Simply calculate the values of the difference quotient f (a + h) − f (a) h (14) with successively smaller nonzero values of h to see whether a limiting numerical value is apparent. EXAMPLE 5 Find by numerical investigation (an approximation to) the line tangent to the graph of f (x) = x + at the point (2, 52 ). 60 1 x (15) Tangent Lines and Slope Predictors SECTION 2.1 t TEXAS INSTRUMENTS t TI-83 FIGURE 2.1.18 A calculator prepared to calculate f (a + h) − f (a) with h 1 f (x) = x + . x h 0.1 0.01 0.001 0.0001 0.00001 ↓ 0 f (2 + h) − f (2) h 0.76190 0.75124 0.75012 0.75001 0.75000 ↓ 3 4 FIGURE 2.1.21 Numerical investigation of the limit in (13) with 1 f (x) = x + , a = 2. x TEXAS INSTRUMENTS TI-83 48SX FIGURE 2.1.19 Approximating f (a + h) − f (a) lim . h→0 h 61 SCIENTIFIC EXPANDABLE FIGURE 2.1.20 A calculator prepared to compute f (x + h) − f (x) . h Solution Figure 2.1.18 shows a TI calculator prepared to calculate the difference quotient in Eq. (14) with the function f in Eq. (15). As indicated in Fig. 2.1.19, successive values of this quotient can then be calculated by brief “one-liners.” Figure 2.1.20 shows an HP calculator prepared to define the same quotient; then evaluation of the expression 'M(2,0.0001)' yields the approximate value m ≈ 0.75001. In this way we get the table shown in Fig. 2.1.21, which suggests that the slope of the line tangent to the graph of f (x) at the point (2, 52 ) is m = 34 . If so, then the tangent line at this point has the point-slope equation y− 5 2 = 34 (x − 2); y = 34 x + 1. that is, Our numerical investigation does not constitute a rigorous proof that this actually is the desired tangent line, but Figs. 2.1.22 and 2.1.23 showing the computer-generated graphs y=x+ 1 x and y= 3 x +1 4 ◗ are strong evidence that we’ve got it right. (Do you agree?) 3 5 1 y=x+x 4 3 y y 2.5 2 y = 43 x + 1 1 0 0 1 1 y=x+x y = 43 x + 1 2 x 3 4 FIGURE 2.1.22 The curve and its tangent line (Example 5). 5 2 1.5 2 x 2.5 FIGURE 2.1.23 The curve and its tangent line magnified near (2, 52 ). 2.1 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The straight line through (a, a 2 ) and (a + h, (a + h)2 ) has slope 2a + h. 2. The straight line tangent to the graph of f (x) = x 2 at the point (a, a 2 ) has slope 2a. 61 62 CHAPTER 2 Prelude to Calculus 3. The straight line tangent to the graph of y = f (x) at the point (a, f (a)) has slope f (a + h) − f (a) . h 4. The straight line tangent to the parabola f (x) = px 2 + q x + r at the point (a, f (a)) has slope 2 pa + q. 5. If the nonvertical lines L 1 and L 2 have slopes m 1 and m 2 , respectively, and L 1 and L 2 are perpendicular, then m 1 m 2 = 1. 6. Every horizontal line has slope zero. 7. To find the highest point on the graph of y = A(x) = − 35 x 2 + 18x, find where the line tangent to the graph has slope zero. 8. The slope-predictor for A(x) = − 35 x 2 + 18x is m(x) = 2x + 18. 9. An equation of the straight line tangent to the graph of y = 2x 2 − 3x + 5 at the point (−1, 10) is y = 3x − 7. 10. Example 5 shows how to find the slope-predictor for the function f (x) = x 3 . 2.1 CONCEPTS: QUESTIONS AND DISCUSSION 1. What is the slope-predictor function for the straight line with equation y = 17x − 21? 2. Can two different parabolas with equations of the form y = px 2 + q x + r have the same slope-predictor function? 3. The vertex of the parabola with equation y = px 2 + q x + r is its highest point (if p < 0) or its lowest point (if p > 0). As indicated in Fig. 2.1.17, it is apparent that this vertex is the single point of the parabola at which the tangent line is horizontal. Is it true—for any given curve y = f (x)—that a point on the graph at which the tangent line is horizontal is either the highest or the lowest point on the graph? 2.1 PROBLEMS In Problems 1 through 14, first apply the slope-predictor formula in (10) for quadratic functions to write the slope m(a) of the line tangent to y = f (x) at the point where x = a. Then write an equation of the line tangent to the graph of f at the point (2, f (2)). 1. f (x) ≡ 5 2. f (x) = x 3. f (x) = x 2 4. f (x) = 1 − 2x 2 5. f (x) = 4x − 5 6. f (x) = 7 − 3x 7. f (x) = 2x 2 − 3x + 4 8. f (x) = 5 − 3x − x 2 9. f (x) = 2x(x + 3) x 2 11. f (x) = 2x − 10 13. f (x) = (2x + 1)2 − 4x 10. f (x) = 3x(5 − x) In Problems 25 through 35, use the slope-predictor formula for quadratic functions as necessary. In Problems 25 through 27, write equations for both the line tangent to, and the line normal to, the curve y = f (x) at the given point P. 25. y = x 2 ; P(−2, 4) 26. y = 5 − x − 2x 2 ; P(−1, 4) 27. y = 2x 2 + 3x − 5; P(2, 9) 28. Prove that the line tangent to the parabola y = x 2 at the point (x0 , y0 ) intersects the x-axis at the point (x0 /2, 0). See Fig. 2.1.24. 12. f (x) = 4 − (3x + 2)2 14. f (x) = (2x+3)2 −(2x−3)2 y y = x2 In Problems 15 through 24, find all points of the curve y = f (x) at which the tangent line is horizontal. 15. y = 10 − x 2 16. y = 10x − x 2 17. y = x − 2x + 1 x 2 19. y = x − 10 21. y = (x + 3)(x − 5) 18. y = x + x − 2 2 23. y = 70x − x 2 62 2 20. y = x(100 − x) 22. y = (x − 5)2 x 2 24. y = 100 1 − 10 (x0, y0) x ( 20, 0) FIGURE 2.1.24 The parabola and tangent line of Problem 28. x Tangent Lines and Slope Predictors SECTION 2.1 29. If a ball is thrown straight upward with initial velocity 96 ft/s, then its height t seconds later is y(t) = 96t − 16t 2 feet. Determine the maximum height the ball attains by finding the point in the parabola y(t) = 96t − 16t 2 where the tangent line is horizontal. 30. According to Problem 40 of Section 1.1, the area of a rectangle with base of length x and perimeter 100 is A(x) = x(50 − x). Find the maximum possible area of this rectangle by finding the point on the parabola A = x(50 − x) at which the tangent line is horizontal. 31. Find the maximum possible value of the product of two positive numbers whose sum is 50. 32. Suppose that a projectile is fired at an angle of 45◦ from the horizontal. Its initial position√is the origin in the x y-plane, and its initial velocity is 100 2 ft/s (Fig. 2.1.25). Then its trajectory will be the part of the parabola y = x −(x/25)2 for which y 0. (a) How far does the projectile travel (horizontally) before it hits the ground? (b) What is the maximum height above the ground that the projectile attains? y = x 2 closest to the point (3, 0). It’s now time for you to find that point. (Suggestion: Draw a figure like Fig. 2.1.26. The cubic equation you should obtain has one solution that is apparent by inspection.) Let P(a, f (a)) be a fixed point on the graph of y = f (x). If h > 0, then Q(a + h, f (a + h)) lies to the right, and R(a − h, f (a − h)) lies to the left, of P. Does Fig. 2.1.27 make it appear plausible—for h > 0 and h very small—that the slope m RQ = 1 f (a + h) − f (a − h) = (m PQ + m RP ) 2h 2 is generally an especially good approximation to the slope m of the line tangent to the graph at P? In particular, the “symmetric difference quotient” m R Q is generally a better approximation to m than either the standard right-hand difference quotient mPQ = mRP = 2 ( 25x ) x f (a + h) − f (a) h or the left-hand difference quotient y y=x− 63 f (a) − f (a − h) . h In Problems 36 through 48, use a calculator or computer to investigate numerically the slope m of the line tangent to the given graph at P(a, f (a)) by calculating both m P Q and m R Q for h = 0.1, 0.01, 0.001, . . . . Check the resulting value of m by plotting both the graph of y = f (x) and the alleged tangent line. FIGURE 2.1.25 The trajectory of the projectile of Problem 32. y = f (x) y 33. One of the two lines that pass through the point (3, 0) and are tangent to the parabola y = x 2 is the x-axis. Find an equation for the other line. (Suggestion: First find the value of the number a shown in Fig. 2.1.26.) Q(a + h, f (a + h)) R(a − h, f(a − h)) y P(a, f (a)) y = x2 a−h (a, a2 ) (3, 0) a a+h x FIGURE 2.1.27 Three different approximations to the slope of a tangent line. x FIGURE 2.1.26 Two lines tangent to the parabola of Problem 33. 34. Write equations for the two straight lines that pass through the point (2, 5) and are tangent to the parabola y = 4x − x 2 . (Suggestion: Draw a figure like Fig. 2.1.26.) 35. Between Examples 3 and 4 we raised—but did not answer— the question of how to locate the point on the graph of 36. f (x) = x 2 ; a = −1 38. f (x) = x 3 ; a = −1 √ 40. f (x) = x; a = 4 1 ; a = − 12 x 44. f (x) = sin 10π x; a = 0 42. f (x) = 1 46. f (x) = sin 10π x; a = 20 √ 48. f (x) = 25 − x 2 ; a = 3 37. f (x) = x 3 ; a = 2 √ 39. f (x) = x; a = 1 1 41. f (x) = ; a = 1 x 43. f (x) = cos x; a = 0 45. f (x) = cos x; a = 14 π √ 47. f (x) = 25 − x 2 ; a = 0 63 64 CHAPTER 2 Prelude to Calculus 2.1 INVESTIGATION: Numerical Slope Approximations 4 3 (x2, y2) y 2 P 1 (x1, y1) 0 In each of the problems listed below, it is known that the slope m of the tangent line to the graph y = f (x) at the fixed point P(a, f (a)) is either an integer or the reciprocal of a single-digit integer. Use this fact to determine m numerically by using a graphing facility (calculator or computer) with which you can “grab” the xy-coordinates of a selected point on the graph. Suppose you “zoom in” on the point P, and at the kth zoom record the coordinates (x1 , y1 ) and (x2 , y2 ) of two points located on either side of P (as indicated in Fig. 2.1.28). Then you can approximate the value of m by calculating the value of the difference quotient y = f(x) 0 1 2 mk = 3 x FIGURE 2.1.28 Points on either side of P. y2 − y1 y = . x x2 − x1 After enough zooms, it should be clear what rational value the approximate slopes m 1 , m 2 , . . . , are approaching. 1. f (x) = x 2 ; P = P(−2, 4); m(−2) = ? √ 2. f (x) = x; P = P(1, 1); m(1) = ? 1 3. f (x) = ; P = P(2, 1/2); m(2) = ? x 12 4. f (x) = 2 ; P = P(−4, 3/4); m(−4) = ? x √ 5. f (x) = x 2 − 9; P = P(5, 4); m(5) = ? √ π 3 6. f (x) = sin ; P = P(3, 3 3/2π ); m(3) = ? π x 2.2 THE LIMIT CONCEPT In Section 2.1 we defined the slope m of the line tangent to the graph y = f (x) at the point P(a, f (a)) to be m = lim h→0 f (a + h) − f (a) . h (1) The graph that motivated this definition is repeated in Fig. 2.2.1, with a + h relabeled as x (so that h = x − a). We see that x approaches a as h approaches zero, so Eq. (1) can be written in the form m = lim x→a f (x) − f (a) . x −a y (2) y = f (x) Q(x, f (x)) f(x) − f (a) = f(a + h) − f (a) P(a, f (a)) x−a=h a x=a+h FIGURE 2.2.1 The slope m at P(a, f (a)) can be f (x) − f (a) . defined in this way: m = lim x→a x −a 64 x The Limit Concept SECTION 2.2 65 Thus the computation of m amounts to the determination of the limit, as x approaches a, of the function f (x) − f (a) . (3) x −a In order to develop general methods for calculating such limits, we need to investigate more fully the meaning of the statement g(x) = lim f (x) = L . x→a (4) This is read “the limit of f (x) as x approaches a is L.” We sometimes write Eq. (4) in the concise form f (x) → L as x → a. The function f need not be defined at the point x = a in order for us to discuss the limit of f at a. The actual value of f (a)—if any—actually is immaterial. It suffices for f (x) to be defined for all points other than a in some neighborhood of a—that is, for all x = a is some open interval containing a. This is exactly the situation for the function in Eq. (3), which is defined except at a (where the denominator is zero). The following statement presents the meaning of Eq. (4) in intuitive language. y y = f(x) Idea of the Limit We say that the number L is the limit of f (x) as x approaches a provided that we can make the number f (x) as close to L as we please merely by choosing x sufficiently near, though not equal to, the number a. (x, f (x)) L (a, L) a x FIGURE 2.2.2 Graphical interpretation of the limit concept. x What this means, roughly, is that f (x) tends to get closer and closer to L as x gets closer and closer to a. Once we decide how close to L we want f (x) to be, it is necessary that f (x) be that close to L for all x sufficiently close to (but not equal to) a. Figure 2.2.2 shows a graphical interpretation of the limit concept. As x approaches a (from either side), the point (x, f (x)) on the graph y = f (x) must approach the point (a, L). In this section we explore the idea of the limit, mainly through the investigation of specific examples. A precise statement of the definition of the limit appears in Section 2.3. EXAMPLE 1 Investigate the value of lim x→3 x −1 . x +2 Investigation This is an investigation (rather than a solution) because numerical calculations may strongly suggest the value of a limit but cannot establish its value with certainty. The table in Fig. 2.2.3 gives values of x −1 , x +2 correct to six rounded decimal places, for values of x that approach 3 (but are not equal to 3). The first and third columns of the table show values of x that approach 3 both from the left and from the right. Now examine the table—read down the columns for x, because down is the table’s direction for “approaches”—to see what happens to the corresponding values of f (x). The data clearly suggest that f (x) = lim x→3 x −1 2 = . x +2 5 ◗ REMARK 1 The graph of f (x) = (x − 1)/(x + 2) in Fig. 2.2.4 reinforces our guess that f (x) is near 25 when x is near 3. For still more reinforcement you can use a graphing calculator or computer to zoom in on the point on the graph where x = 3. 65 66 CHAPTER 2 Prelude to Calculus x x −1 x +2 x x −1 x +2 2 2.9 2.99 2.999 2.9999 ↓ 3 0.250000 0.387755 0.398798 0.399880 0.399988 ↓ 0.4 4 3.1 3.01 3.001 3.0001 ↓ 3 0.500000 0.411765 0.401198 0.400120 0.400012 ↓ 0.4 1 y= x−1 x+2 0.5 y = 0.4 y 0 x=3 -0.5 -1 0 FIGURE 2.2.3 Investigating the limit in Example 1. 1 2 3 x 4 5 6 FIGURE 2.2.4 The limit in Example 1. REMARK 2 Note that we did not simply substitute the value x = 3 into the function f (x) = (x − 1)/(x + 2) to obtain the apparent value 25 = 0.4 of the limit. Although such substitution would produce the correct answer in this particular case, in many limits it produces either an incorrect answer or no answer at all. (See Examples 2 and 3 and Problems 19 through 36 and 47 through 56.) EXAMPLE 2 Investigate the value of lim x→2 x2 − 4 . x2 + x − 6 Investigation The numerical data shown in Fig. 2.2.5 certainly suggest that lim x→2 x 1 1.5 1.9 1.99 1.999 1.9999 ↓ 2 x2 4 x2 − 4 = . 2 x +x −6 5 x2 − 4 +x −6 0.750000 0.777778 0.795918 0.799599 0.799960 0.799996 ↓ 0.8 x 3 2.5 2.1 2.01 2.001 2.0001 ↓ 2 ◗ x2 x2 − 4 +x −6 0.833333 0.818182 0.803922 0.800399 0.800040 0.800004 ↓ 0.8 FIGURE 2.2.5 Investigating the limit in Example 2. REMARK The function 1.5 (2, 1.2) 1 y = 0.8 y y= x2 − 4 x2 + x − 6 0.5 0 0 0.5 1 1.5 2 2.5 3 3.5 4 x FIGURE 2.2.6 The limit in Example 2. 66 f (x) = x2 − 4 x2 + x − 6 is not defined at x = 2, so we cannot merely substitute 2 for x. But if we let ⎧ x2 − 4 ⎪ ⎪ if x = 2, ⎨ 2 x +x −6 g(x) = ⎪ 6 ⎪ ⎩ if x = 2, 5 then g(x) is defined at x = 2 (and agrees with f (x) elsewhere). Is it clear to you that f and g must have the same limit at x = 2? Figure 2.2.6 shows the graph y = g(x), including the isolated point (2, 1.2) on its graph. The Limit Concept SECTION 2.2 67 √ t 1.0 0.5 0.1 0.05 0.01 0.005 ↓ 0 t + 25 − 5 t 0.099020 0.099505 0.099900 0.099950 0.099990 0.099995 ↓ 0.1 FIGURE 2.2.7 Investigating the limit in Example 3. √ t + 25 − 5 EXAMPLE 3 Investigate the value of lim . t→0 t Investigation Here we cannot make a guess by substituting t = 0 because the fraction √ t + 25 − 5 g(t) = t is not defined when t = 0. But the numerical data shown in Fig. 2.2.7 indicate that √ t + 25 − 5 1 = . lim t→0 t 10 1 We can attempt to corroborate this result graphically by zooming in on the point (1, 10 ). The plot shown in Fig. 2.2.8 does not contradict the indicated limit, but somehow is 1 for unconvincing because it “goes too far” and suggests (incorrectly!) that g(t) = 10 t = 0. The problem is that the scale on the y-axis is too coarse. The magnification 1 . ◗ shown in Fig. 2.2.9 does appear to substantiate the limiting value of 10 1 0.101 0.5 y= 0.1005 t + 25 − 5 t y y 0 (0, 1/10) 0.1 y= -0.5 -1 -1 t + 25 − 5 t 0.0995 -0.5 0 t 0.5 0.099 -1 1 FIGURE√2.2.8 Graph of t + 25 − 5 for g(t) = t −1 t + 1, −1 y + 1. -0.5 0 t 0.5 1 FIGURE√2.2.9 Graph of t + 25 − 5 for −1 t + 1, g(t) = t 0.099 y 0.101. REMARK Can you see that, upon dividing each number in the second column of Fig. 2.2.7 by 10000, one might well suspect that √ t + 25 − 5 = 0? (Wrong!) lim t→0 10000t In fact, the value of this limit (as we will see in Example 13) is exactly 10−5 = 0.00001, not zero. This fact constitutes a warning that numerical investigations of limits are not conclusive. The numerical investigation in Example 3 is incomplete because the table in Fig. 2.2.7 shows values of the function g(t) on only one side of the point t = 0. But in order that limx→a f (x) = L, it is necessary for f (x) to approach L both as x approaches a from the left and as x approaches a from the right. If f (x) approaches different values as x approaches a from different sides, then limx→a f (x) does not exist. In Section 2.3 we discuss such one-sided limits in more detail. EXAMPLE 4 Investigate lim f (x), given x→0 f (x) = x 1 = −1 |x| if x > 0, if x < 0. 67 68 CHAPTER 2 Prelude to Calculus y +1 x −1 FIGURE 2.2.10 The graph of x f (x) = (Example 4). |x| In Example 5 the value obtained by substituting x = a in F(x) to find limx→a F(x) is incorrect. EXAMPLE 5 Evaluate lim F(x) where y F(0) = 0 FIGURE 2.2.11 The graph of the function F of Example 5. Solution From the graph of f shown in Fig. 2.2.10, it is apparent that f (x) → 1 as x → 0 from the right and that f (x) → −1 as x → 0 from the left. In particular, there are positive values of x as close to zero as we please such that f (x) = 1 and negative values of x equally close to zero such that f (x) = −1. Hence we cannot make f (x) as close as we please to any single value of L merely by choosing x sufficiently close to zero. Therefore, x does not exist. ◗ lim x→0 |x| x→0 F(x) = x if x = 0, if x = 0. 1 0 The graph of F is shown in Fig. 2.2.11. Solution The fact that F(x) = 1 for every value of x = 0 in any neighborhood of zero implies that lim F(x) = 1. x→0 But note that the value of the limit at x = 0 is not equal to the functional value F(0) = 0 there. ◗ The Limit Laws Numerical investigations such as those in Examples 1 through 3 provide us with an intuitive feeling for limits and typically suggest the correct value of a limit. But most limit computations are based neither on merely suggestive (and imprecise) numerical estimates nor on direct (but difficult) applications of the definition of limit. Instead, such computations are performed most easily and naturally with the aid of the limit laws that we give next. These “laws” actually are theorems, whose proofs (based on the precise definition of the limit) are included in Appendix D. Constant Law If f (x) ≡ C, where C is a constant [so f (x) is a constant function], then lim f (x) = lim C = C. x→a x→a (5) Sum Law If both of the limits lim f (x) = L x→a and lim g(x) = M x→a exist, then lim [ f (x) ± g(x)] = x→a lim f (x) ± lim g(x) = L ± M. x→a x→a (6) (The limit of a sum is the sum of the limits; the limit of a difference is the difference of the limits.) 68 The Limit Concept SECTION 2.2 69 Product Law If both of the limits lim f (x) = L lim g(x) = M and x→a x→a exist, then lim [ f (x)g(x)] = lim f (x) x→a lim g(x) = L M. x→a x→a (7) (The limit of a product is the product of the limits.) Quotient Law If both of the limits lim f (x) = L x→a lim g(x) = M and x→a exist and if M = 0, then lim x→a lim f (x) L f (x) x→a = = . g(x) lim g(x) M (8) x→a (The limit of a quotient is the quotient of the limits, provided that the limit of the denominator is not zero.) Root Law If n is a positive integer and if a > 0 for even values of n, then √ √ lim n x = n a. x→a (9) The case n = 1 of the root law is obvious: lim x = a. (10) x→a Examples 6 and 7 show how the limit laws can be used to evaluate limits of polynomials and rational functions. EXAMPLE 6 lim x 2 + lim 2x + lim 4 x→3 x→3 x→3 2 = lim x + 2 lim x + lim 4 = 32 + 2 · 3 + 4 = 19. lim (x 2 + 2x + 4) = x→3 x→3 x→3 x→3 ◗ EXAMPLE 7 lim x→3 lim (2x + 5) 2x + 5 x→3 = x 2 + 2x + 4 lim (x 2 + 2x + 4) x→3 = 32 11 2·3+5 = . +2·3+4 19 ◗ NOTE In Examples 6 and 7, we systematically applied the limit laws until we could simply substitute 3 for limx→3 x at the final step. To determine the limit of a quotient of polynomials, we must verify before this final step that the limit of the denominator is not zero. If the denominator limit is zero, then the limit may fail to exist. 1 . x→1 (x − 1)2 EXAMPLE 8 Investigate lim 69 70 CHAPTER 2 Prelude to Calculus y y= 1 (x − 1)2 y=L+∋ L y=L−∋ Solution Because limx→1 (x − 1)2 = 0, we cannot apply the quotient law. Moreover, we can make 1/(x − 1)2 arbitrarily large by choosing x sufficiently close to 1. Hence 1/(x − 1)2 cannot approach any (finite) number L as x approaches 1. Therefore, the limit in this example does not exist. You can see the geometric reason if you examine the graph of y = 1/(x − 1)2 in Fig. 2.2.12. As x → 1, the corresponding point (x, y) ascends the curve near the vertical line x = 1. It must therefore leave the indicated strip between the two horizontal lines x = L − and x = L + that bracket the proposed limit L. Thus, the point (x, y) cannot approach the point (1, L) as x → 1. ◗ x x=1 FIGURE 2.2.12 The graph of 1 (Example 8). y= (x − 1)2 EXAMPLE 9 Investigate lim x→2 x2 − 4 . x2 + x − 6 Solution We cannot immediately apply the quotient law (as we did in Example 7) because the denominator approaches zero as x approaches 2. If the numerator were approaching some number other than zero as x → 2, then the limit would fail to exist (as in Example 8). But here the numerator does approach zero, so there is a possibility that a factor of the numerator can be canceled with the same factor of the denominator, thus removing the zero-denominator problem. Indeed, lim x→2 x2 − 4 (x − 2)(x + 2) = lim x 2 + x − 6 x→2 (x − 2)(x + 3) 4 x +2 = . = lim x→2 x + 3 5 We can cancel the factor x − 2 because it is nonzero: x = 2 when we evaluate the limit as x approaches 2. Moreover, this verifies the numerical limit of 0.8 that we found in ◗ Example 2. Substitution of Limits It is tempting to write lim x→−4 x2 + 9 = = lim (x 2 + 9) x→−4 (−4)2 + 9 = √ 25 = 5. (11) But can we simply “move the limit inside the radical” in Eq. (11)? To analyze this question, let us write f (x) = √ x and g(x) = x 2 + 9. Then the function that appears in Eq. (11) is the composite function f (g(x)) = g(x) = x 2 + 9. (Recall that the left-hand expression in this equation is read “ f of g of x.”) Hence our question is whether or not lim f (g(x)) = f lim g(x) . x→a x→a The next limit law answers this question in the affirmative, provided that the “outside function” f meets a certain condition; if so, then the limit of the composite function f (g(x)) as x → a may be found by substituting into the function f the limit of g(x) as x → a. 70 The Limit Concept SECTION 2.2 71 Substitution Law Limits of Compositions Suppose that lim g(x) = L and that lim f (x) = f (L). x→a x→L Then lim f (g(x)) = f x→a lim g(x) = f (L). x→a (12) Thus the condition under which Eq. (12) holds is that the limit of the outer function f not only exists at x = L, but also is equal to the “expected” value of f —namely, f (L). In particular, because √ √ lim (x 2 + 9) = 25 and lim x = 25 = 5, x→−4 x→25 this condition is satisfied in Eq. (11). Hence the computations shown there are valid. In this section we use only the following special case of the substitution law. With f (x) = x 1/n , where n is a positive integer, Eq. (12) takes the form lim √ n x→a g(x) = n lim g(x), (13) x→a under the assumption that the limit of g(x) exists as x → a (and is positive if n is even). With g(x) = x m , where m is a positive integer, Eq. (13) in turn yields lim x m/n = a m/n , (14) x→a with the condition that a > 0 if n is even. Equations (13) and (14) may be regarded as generalized root laws. Example 10 illustrates the use of these special cases of the substitution law. EXAMPLE 10 √ √ √ 3 3 lim 3 x + 20 x = 3 lim (3x 3/2 + 20 x ) x→4 x→4 √ 1/3 lim 3x 3/2 + lim 20 x x→4 x→4 √ 1/3 3/2 = 3 · 4 + 20 4 √ 3 = (24 + 40)1/3 = 64 = 4. = [using Eq. (13)] [using the sum law] [using Eq. (14)] ◗ Slope-Predictor Functions Our discussion of limits began with the slope m a = lim h→0 f (a + h) − f (a) h (15) of the line tangent to the graph y = f (x) at the point (a, f (a)). The lines tangent to y = f (x) at different points have different slopes. Thus if we replace a with x in Eq. (15), we get a new function defined by m(x) = lim h→0 f (x + h) − f (x) . h (16) This function m may be regarded as a “slope predictor” for lines tangent to the graph y = f (x). It is a new function derived from the original function f (x), and in Chapter 3 we will call it the derivative of f . 71 72 CHAPTER 2 Prelude to Calculus 20 y = 2x + 4 10 EXAMPLE 11 In Section 2.1 we saw that the line tangent to the graph y = px 2 + q x + r at the point where x = a has slope m a = 2 pa + q. Hence the slope-predictor function for the quadratic function y 0 f (x) = px 2 + q x + r (17) m(x) = 2 px + q. (18) is the linear function -10 -20 -10 y = x2 + 4x − 12 -5 0 x 5 FIGURE 2.2.13 The parabola y = x 2 + 4x − 12 and its slope predictor m(x) = 2x + 4. 10 Figure 2.2.13 illustrates the case p = 1, q = 4, r = −12. It is worth noting that the x-intercept where m(x) = 0 corresponds to the point of the parabola y = f (x) where ◗ the tangent line is horizontal. The slope-predictor definition in Eq. (16) calls for us to carry out the following four steps. 1. 2. 3. 4. Write the definition of m(x). Substitute into this definition the formula of the given function f . Make algebraic simplifications until Step 4 can be carried out. Determine the value of the limit as h → 0. Note that x may be thought of as a constant throughout this computation—it is h that is the variable in this four-step process. EXAMPLE 12 Find the slope-predictor function for the function 1 x that was investigated numerically in Example 5 of Section 2.1. f (x) = x + Solution The first two steps in the preceding list yield 1 1 x +h+ − x+ f (x + h) − f (x) x +h x = lim . m(x) = lim h→0 h→0 h h We cancel the two copies of x in the numerator and proceed to simplify algebraically, first finding a common denominator in the numerator: 1 1 h+ − x +h x m(x) = lim h→0 h h(x + h)x + x − (x + h) = lim h→0 h(x + h)x h(x + h)x − h . = lim h→0 h(x + h)x Now we can divide numerator and denominator by h (because h = 0) and finally apply the sum, product, and quotient laws to evaluate the limit as h → 0: h(x + h)x − h h→0 h(x + h)x (x + h)x − 1 x2 − 1 1 = lim = = 1 − 2. h→0 (x + h)x x2 x m(x) = lim For instance, the slope of the line tangent to y=x+ at the point (2, 52 ) is m(2) = 72 3 4 1 x (thus verifying the result in Example 5 of Section 2.1). ◗ The Limit Concept SECTION 2.2 73 Example 13 illustrates an algebraic procedure often used in “preparing” functions before taking limits. This procedure can be applied when roots are present and resembles the simple computation √ √ 5+ 2 1 1 √ =√ √ ·√ √ √ 5− 2 5− 2 5+ 2 √ √ √ √ 5+ 2 5+ 2 = . = 5−2 3 √ EXAMPLE 13 Find the slope-predictor function for the function f (x) = x . Solution √ m(x) = lim h→0 x +h − h √ x . (19) To prepare the fraction for evaluation of the √ √ limit, we first multiply the numerator and denominator by the conjugate x + h + x of the numerator: √ √ √ √ x +h − x x +h + x m(x) = lim ·√ √ h→0 h x +h + x (x + h) − x = lim √ √ h→0 h( x + h + x ) 1 = lim √ √ . h→0 x +h + x Thus 1 m(x) = √ . 2 x (20) (In the final step we used the sum, quotient, and root laws—we did not simply substi◗ tute 0 for h.) Note that if we equate the right-hand sides of Eqs. (19) and (20) and take x = 25, then we get the limit in Example 3: √ 25 + h − 5 1 = . lim h→0 h 10 (The t in Example 3 has been replaced here with h.) And if we divide both sides by 10000 we find that √ 25 + h − 5 1 = = 0.00001, lim h→0 10000h 100000 as claimed in the remark following Example 3. 2.2 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Suppose that the function f is given together with the point P(a, f (a)) on its graph. Then the slope of the straight line tangent to the graph of f at the point P is f (x) − f (a) . g(x) = x −a 2. Suppose that the function f is given together with the point P(a, f (a)) on its graph. Then the slope of the straight line tangent to the graph of f at the point P is the limiting value, as x approaches a, of the function g(x) defined in the preceding item. 73 74 CHAPTER 2 Prelude to Calculus 3. To say that lim g(x) = L means that g(x) can be made arbitrarily close to the x→a number L merely by ensuring that x is sufficiently close to (but not equal to) the number a. 4. lim x 3 = 6. x→2 x2 − 4 , then lim g(x) = g(2). x→2 x2 + x − 6 x 6. If f (x) = , then lim f (x) does not exist because f (0) is undefined. x→0 |x| 1 if x = 0, 7. If F(x) = then lim F(x) = 0. x→0 0 if x = 0, 5. If g(x) = 8. If lim f (x) = L and lim g(x) = M, then lim f (x) · g(x) = L · M. x→a x→a x→a 9. If lim f (x) = L and lim g(x) = M, then lim x→a x→a x→a 10. It follows from the limit laws that lim x→3 L f (x) = . g(x) M 25 − x 2 = 4. 2.2 CONCEPTS: QUESTIONS AND DISCUSSION 1. The sum, product, and quotient laws imply that if the limits lim f (x) and lim g(x) x→a (21) x→a both exist, then the limit lim [ f (x) g(x)] x→a (22) also exists—with the symbol denoting either +, −, ×, or ÷ (and assuming in the case of division that limx→a g(x) = 0). Can you produce examples— in all four cases—of functions such that neither of the limits in (21) exists, but nevertheless the limit in (22) does exist? It may help to review the examples of nonexisting limits in this section. 2. Can you produce examples of functions f and g such that both lim g(x) = b x→a lim f (x) = c and exist, but lim f (g(x)) = f x→a x→b lim g(x) ? x→a If so, why does this not contradict the substitution law of limits? 2.2 PROBLEMS Apply the limit laws of this section to evaluate the limits in Problems 1 through 18. Justify each step by citing the appropriate limit law. (x 2 + 1)3 x→3 (x 3 − 25)3 √ 9. lim 4x + 5 7. lim x→1 1. lim (3x 2 + 7x − 12) x→3 11. lim (x 2 − 1)3/2 2. lim (x 3 − 3x 2 + 5) x→3 x→−2 3. lim (x 2 − 1)(x 7 + 7x − 4) 13. lim 4. lim (x 3 − 3x + 3)(x 2 + 2x + 5) 15. lim z→8 x→1 x→−2 x +1 5. lim 2 x→1 x + x + 1 74 t +2 6. lim 2 t→−2 t + 4 z 2/3 √ z − 2z (w − 2)4 x +2 17. lim 3 x→−2 (x − 2)2 w→0 (3z 2 + 2z + 1)10 8. lim z→−1 (z 3 + 5)5 √ 10. lim 27 − y y→4 t +8 12. lim t→−4 25 − t 2 14. lim 3t 3 + 4t − 5 3 t→2 16. lim 3 t→−4 18. lim y→5 (t + 1)6 2y 2 + 2y + 4 6y − 3 1/3 The Limit Concept SECTION 2.2 75 In Problems 19 through 28, note first that the numerator and denominator have a common algebraic factor (as in Example 9). Use this fact to help evaluate the given limit. x +1 19. lim 2 x→−1 x − x − 2 x2 + x − 2 21. lim 2 x→1 x − 4x + 3 t2 − 9 20. lim t→3 t − 3 4y 2 − 1 22. lim 2 y→−1/2 4y + 8y + 3 t 2 + 6t + 9 t→−3 t2 − 9 (z + 2)2 25. lim 4 z→−2 z − 16 x3 − 1 27. lim 4 x→1 x − 1 x2 − 4 − 2x − 8 t 3 − 9t 26. lim 2 t→3 t − 9 y 3 + 27 28. lim 2 y→−3 y − 9 23. lim 24. lim x→2 3x 2 In Problems 29 through 36, evaluate those limits that exist. 1 1 − x 3 29. lim x→3 x − 3 x −4 31. lim √ x→4 x −2 √ t +4−2 33. lim t→0 t 35. lim x→4 x 2 − 16 √ 2− x 30. 32. 34. 36. 1 1 − 2 + t 2 lim t→0 t √ 3− x lim x→9 9 − x 1 1 1 lim − √ h→0 h 3 9+h √ √ 1+x − 1−x lim x→0 x In Problems 37 through 46, use the four-step process illustrated in Examples 12 and 13 to find a slope-predictor function for the given function f (x). Then write an equation for the line tangent to the curve y = f (x) at the point where x = 2. 1 x 37. f (x) = x 3 38. f (x) = 1 x2 2 41. f (x) = x −1 1 43. f (x) = √ x +2 √ 45. f (x) = 2x + 5 1 x +1 x 42. f (x) = x −1 3 44. f (x) = x 2 + x 39. f (x) = (1 + x)2 − 1 x→0 x √ x +9−3 lim x→0 x 1 1 2 − lim x→0 x (2 + x)3 4 sin x lim x→0 x x − sin x lim x→0 x3 49. 51. 53. 55. lim (1 + x)1/x x→0 is the famous irrational number e (of Chapter 3), whose three-place decimal approximation is e ≈ 2.718. Numerically investigate this limit to approximate e accurate to five decimal places. Corroborate this value graphically by zooming in on the y-intercept of the curve y = (1 + x)1/x . 58. Verify graphically the limit lim x→0 46. f (x) = x2 x +1 48. lim x→1 x4 − 1 x −1 x 3/2 − 8 x→4 x − 4 (3 + x)−1 − (3 − x)−1 52. lim x→0 x 1 − cos x 54. lim x→0 x2 1 x 56. lim 1 + x→0 |x| 50. lim sin x x of Problem 53 by zooming in on the y-intercept of the curve y = (sin x)/x. 59. Investigate the limit lim x→0 x − tan x x3 both numerically and graphically. Determine its value accurate to four decimal places. 60. The value of lim x→0 sin 2x tan 5x is the ratio of two single-digit integers. Determine this value both numerically and graphically. 61. Calculate the value of f (x) = sin 40. f (x) = In Problems 47 through 56, the actual value of the given limit limx→a f (x) is a rational number that is a ratio of two singledigit integers. Guess this limit on the basis of a numerical investigation in which you calculate f (x) for x = a±0.1, x = a±0.05, x = a ±0.01, x = a ±0.005, and so on. Use other similar values of x near a as you wish. 47. lim 57. In contrast with the rational-valued limits in Problems 47 through 56, the value of the limit for x = 12 , 14 , 18 , the value of 1 , 16 π x . . . . What do you now conjecture to be lim sin x→0 π ? x Next calculate f (x) for x = 23 , 29 , you conclude? 2 , 2, 27 81 . . . . Now what do 62. To investigate the limit of f (x) = sin x + 10−5 cos x as x → 0, set your graphing calculator or computer to display exactly four digits to the right of the decimal point. After calculating f (x) with x = 0.1, 0.001, 0.00001, 0.0000001, . . . , what do you conclude? (Your answer may depend on how your particular calculator works.) Now zoom in on the y-intercept of the curve y = f (x) sufficiently to show that the value of the limit is nonzero. What is it? 63. Investigate numerically or graphically (or both) the value of the limit 1 −1/32 lim log10 . x→0 |x| The actual value of this limit is zero, so you’ll see that your calculator or computer cannot always be believed. 75 76 CHAPTER 2 Prelude to Calculus results substantiate the fact that L = ln 10, the value produced by the LN key on your calculator? (b) Show that the slope-predictor function for lines tangent to the graph y = 10x is m(x) = L · 10x . Corroborate this fact by using a calculator or computer to plot the graph of y = 10x and its predicted tangent lines at several different points. 64. (a) Show that the slope of the line tangent to the graph of y = 10x at the point (0, 1) is the number 10h − 1 . h→0 h Investigate this limit numerically and graphically. Do your L = lim 2.2 INVESTIGATION: Limits, Slopes, and Logarithms 8 Generalize the result in Problem 64 of this section. First refer to Fig. 2.2.14. Then suppose that a is a positive constant. Show that the slope of the line tangent to the graph of y = a x at the point (0, 1) is the number y = ax 6 4 y 2 ah − 1 . h→0 h L(a) = lim (0, 1) 0 -2 y = L(a)x + 1 -4 -1 -0.5 0 x 0.5 1 (Note how the notation of functions is used in Eq. (1) to emphasize the dependence of the slope on the base constant a.) Next choose at random a pair of positive integers a and b and investigate the numerical values of L(a), L(b), and L(ab). Are your results consistent with the fact that L(ab) = L(a) + L(b), FIGURE 2.2.14 The graph of y = a x and its tangent line at the point (0, 1). (1) (2) in analogy with the law of logarithms log ab = log a + log b? (3) At this point the connection between Eqs. (2) and (3) is surely an enigma rather than an explanation. The mystery will be explained in Section 3.8, in which we study natural logarithms. For now, use the LN key on your calculator to find ln a, ln b, and ln ab; compare these with your earlier values of L(a), L(b), and L(ab). You can also follow up these investigations with a computer algebra system: Use it to attempt to evaluate the limit in Eq. (1) symbolically, and then compare the symbolic result with your numerical results. 2.3 MORE ABOUT LIMITS To investigate limits of trigonometric functions, we begin with Fig. 2.3.1, which shows an angle θ with its vertex at the origin, its initial side along the positive x-axis, and its terminal side intersecting the unit circle at the point P. By the definition of the sine and cosine functions, the coordinates of P are P(cos θ, sin θ). From geometry we see that, as θ → 0, the point P(cos θ, sin θ ) approaches the point R(1, 0). Hence cos θ → 1 and sin θ → 0 as θ → 0 through positive values. A similar picture gives the same result for negative values of θ, so we see that y x2 + y2 = 1 P(cos θ , sin θ ) θ R(1, 0) FIGURE 2.3.1 An angle θ . 76 x lim cos θ = 1 θ→0 and lim sin θ = 0. θ →0 (1) Equation (1) says simply that the limits of the functions cos θ and sin θ as θ → 0 are equal to their respective values at θ = 0: cos 0 = 1 and sin 0 = 0. The limit of the quotient (sin θ )/θ as θ → 0 plays a special role in the calculus of trigonometric functions. For instance, it is needed to find slopes of lines tangent to trigonometric graphs such as y = cos x and y = sin x. Note that the value of the quotient (sin θ )/θ is not defined when θ = 0. (Why not?) But a calculator set in radian mode provides us with the numerical evidence shown in Fig. 2.3.2. This table strongly suggests that the limit of (sin θ)/θ is 1 as θ → 0. This conclusion is supported by the graph of y = (sin x)/x shown in Fig. 2.3.3, where it appears that the point (x, y) on the curve is near (0, 1) when x is near zero. Later in this section we provide a proof of the following result. More About Limits SECTION 2.3 θ sin θ θ ±1.0 ±0.5 ±0.1 ±0.05 ±0.01 ±0.005 ±0.001 .. . ↓ 0 0.84147 0.95885 0.99833 0.99958 0.99998 1.00000 1.00000 .. . ↓ 1 2 1 y (0, 1) y = (sin x)/x 0 -1 -2 -10 -5 0 x FIGURE 2.3.3 y = FIGURE 2.3.2 The numerical data sin θ = 1. suggest that lim θ →0 θ THEOREM 1 77 x = 0. 5 10 sin x for x The Basic Trigonometric Limit lim x→0 sin x = 1. x (2) As in Examples 1 and 2, many other trigonometric limits can be reduced to the one in Theorem 1. EXAMPLE 1 Show that 1 − cos x = 0. x→0 x lim (3) Solution We multiply the numerator and denominator in Eq. (3) by the “conjugate” 1 + cos x of the numerator 1 − cos x. Then we apply the identity 1 − cos2 x = sin2 x. This gives 1 − cos x 1 − cos x 1 + cos x sin2 x = lim · = lim x→0 x→0 x→0 x(1 + cos x) x x 1 + cos x 0 sin x sin x = lim lim =1· = 0. x→0 x x→0 1 + cos x 1+1 lim In the last step we used all the limits in Eqs. (1) and (2). EXAMPLE 2 Evaluate lim x→0 ◗ tan 3x . x Solution tan 3x tan 3x tan θ = 3 lim = 3 lim (θ = 3x) lim x→0 x→0 3x θ→0 θ x sin θ sin θ = 3 lim because tan θ = θ→0 θ cos θ cos θ sin θ 1 lim (by the product law of limits) = 3 lim θ→0 θ θ→0 cos θ 1 = 3 · 1 · = 3. 1 We used the fact that tan θ = (sin θ )/(cos θ ) as well as some of the limits in Eqs. (1) ◗ and (2). 77 78 CHAPTER 2 Prelude to Calculus x π sin x 1 0.5 0.1 0.05 0.01 0.005 0.001 0 0 0 0 0 0 0 Example 3 constitutes a warning: The results of a numerical investigation can be misleading unless they are interpreted with care. EXAMPLE 3 The numerical data shown in the table of Fig. 2.3.4 suggest that the limit lim sin x→0 y = sin(π/x) sin y (4) has the value zero. But it appears in the graph of y = sin(π/x) (for x = 0), shown in Fig. 2.3.5, that the value of sin(π/x) oscillates infinitely often between +1 and −1 as x → 0. Indeed, this fact follows from the periodicity of the sine function, because π/x increases without bound as x → 0. Hence sin(π/x) cannot approach zero (or any other number) as x → 0. Therefore the limit in (4) does not exist. We can explain the potentially misleading results tabulated in Fig. 2.3.4 as follows: Each value of x shown there just happens to be of the form 1/n, the reciprocal of an integer. Therefore, FIGURE 2.3.4 Do you think that π lim sin = 0 (Example 3)? x→0 x 1 π x 0 π π = sin = sin nπ = 0 x 1/n for every nonzero integer n. But with a different selection of “trial values” of x, we might have obtained the results shown in Fig. 2.3.6, which immediately suggest the ◗ nonexistence of the limit in (4). -1 -1 -0.5 0 x 0.5 1 The Squeeze Law of Limits FIGURE 2.3.5 The graph of π y = sin shows infinite oscillation x as x → 0 (Example 3). x 2 9 2 11 2 101 2 103 2 1001 2 1003 sin A final property of limits that we will need is the squeeze law (also known as the “sandwich theorem”). It is related to the fact that taking limits preserves inequalities among functions. Figure 2.3.7 illustrates how and why the squeeze law works and how it got its name. The idea is that g(x) is trapped between f (x) and h(x) near a; both f (x) and h(x) approach the same limit L, so g(x) must approach L as well. A formal proof of the squeeze law can be found in Appendix D. π x +1 Squeeze Law Suppose that f (x) g(x) h(x) for all x = a in some neighborhood of a and also that lim f (x) = L = lim h(x). −1 +1 −1 x→a +1 Then −1 x→a lim g(x) = L x→a FIGURE 2.3.6 Verify the entries in the second column (Example 3). EXAMPLE 4 Figures 2.3.8 and 2.3.9 show two views of the graph of the function g defined for x = 0 by y y = h(x) 1 g(x) = x sin . x y = g(x) L y = f (x) a FIGURE 2.3.7 How the squeeze law works. 78 as well. x As in Example 3, sin(1/x) oscillates infinitely often between +1 and −1 as x → 0. Therefore the graph y = g(x) bounces back and forth between the lines y = +x and y = −x. Because |sin(1/x)| 1 for all x = 0, −|x| x sin 1 +|x| x More About Limits SECTION 2.3 79 2 y = −x 0.2 y=x 0.1 1 y = x sin(1/x) y 0 y 0 -0.1 -1 -2 -2 y=x y = -x -1 0 x 1 -0.2 -0.2 2 FIGURE 2.3.8 The graph of 1 g(x) = x sin for x = 0 x (Example 4). y = x sin (1/x) -0.1 0 x 0.1 0.2 FIGURE 2.3.9 The graph magnified near the origin (Example 4). for all x = 0. Moreover, ±|x| → 0 as x → 0, so with f (x) = −|x| and h(x) = +|x|, it follows from the squeeze law of limits that lim x sin x→0 1 = 0. x (5) ◗ QUESTION Why doesn’t the limit in Eq. (5) follow from the product law of limits with f (x) = x and g(x) = sin(1/x)? One-Sided Limits In Example 4 of Section 2.2 we examined the function y 1 f (x) = x −1 FIGURE 2.3.10 The graph of x again. f (x) = |x| x 1 = −1 |x| if x > 0; if x < 0. The graph of y = f (x) is shown in Fig. 2.3.10. We argued that the limit of f (x) as x → 0 does not exist because f (x) approaches +1 as x approaches zero from the right, whereas f (x) → −1 as x approaches zero from the left. A natural way of describing this situation is to say that at x = 0 the right-hand limit of f (x) is +1 and the left-hand limit of f (x) is −1. Here we define and investigate such one-sided limits. Their definitions will be stated initially in the informal language we used in Section 2.2 to describe the “idea of the limit.” To define the right-hand limit of f (x) at x = a, we must assume that f is defined on an open interval immediately to the right of a. To define the left-hand limit, we must assume that f is defined on an open interval immediately to the left of a. The Right-Hand Limit of a Function Suppose that f is defined on the interval (a, c) immediately to the right of a. Then we say that the number L is the right-hand limit of f (x) as x approaches a (from the right), and we write lim f (x) = L , x→a + (6) provided that we can make the number f (x) as close to L as we please merely by choosing the point x in (a, c) sufficiently close to a. We may describe the right-hand limit in Eq. (6) by saying that f (x) → L as x → a + ; that is, as x approaches a from the right. The symbol a + denotes the righthand, or “positive,” side of the number a (which may be positive, negative, or zero). 79 80 CHAPTER 2 Prelude to Calculus For instance, we see in Fig. 2.3.10 that y lim L x→0+ y = f(x) a x (a) |x| = +1 x (7) because |x|/x is equal to +1 for all x to the right of zero. See Fig. 2.3.11(a) for a more general geometric interpretation of right-hand limits. The Left-Hand Limit of a Function Suppose that f is defined on the interval (c, a) immediately to the left of a. Then we say that the number L is the left-hand limit of f (x) as x approaches a (from the left), and we write (8) lim f (x) = L , y L y = f (x) x→a − a x (b) FIGURE 2.3.11 (a) The right- hand limit of f (x) is L. (b) The left-hand limit of f (x) is L. provided that we can make the number f (x) as close to L as we please merely by choosing the point x in (c, a) sufficiently close to a. We may describe the left-hand limit in Eq. (8) by saying that f (x) → L as x → a − ; that is, as x approaches a from the left. The symbol a − denotes the left-hand or “negative” side of a. For instance, we see in Fig. 2.3.10 that lim x→0− |x| = −1 x (9) because |x|/x is equal to −1 for all x to the left of zero. See Fig. 2.3.11(b) for a more general geometric interpretation of left-hand limits. In Example 4 of Section 2.2 we argued (in essence) that, because the limits in Eqs. (7) and (9) are not equal, the corresponding two-sided limit |x| x does not exist. More generally, Theorem 2 (next) follows from careful consideration of the definitions of all the limits involved. lim x→0 THEOREM 2 One-Sided Limits and Two-Sided Limits Suppose that the function f is defined for x = a in a neighborhood of the point a. Then the two-sided limit lim f (x) x→a exists and is equal to the number L if and only if the one-sided limits y lim f (x) and x→a + 4 3 lim f (x) x→a − both exist and are equal to L. 2 1 −3 −2 −1 1 −1 2 3 4 5 x −2 −3 FIGURE 2.3.12 The graph of the greatest integer function f (x) = [[x]] (Example 5). Theorem 2 is particularly useful in showing that certain (two-sided) limits do not exist, frequently by showing that the left-hand and right-hand limits are not equal to each other. EXAMPLE 5 The graph of the greatest integer function f (x) = [[x]] is shown in Fig. 2.3.12. It should be apparent that if a is not an integer, then lim [[x]] = lim [[x]] = lim [[x]] = [[a]]. x→a + x→a − x→a But if a = n, an integer, then lim [[x]] = n − 1 x→n − and lim [[x]] = n. x→n + Because these left-hand and right-hand limits are not equal, it follows from Theorem 2 ◗ that the limit of f (x) = [[x]] does not exist as x approaches an integer n. 80 More About Limits SECTION 2.3 81 EXAMPLE 6 According to the root law in Section 2.2, √ √ lim x = a if a > 0. √ x→a But the limit of f (x) = x as x → 0− is not defined because the square root of a negative number is undefined. Hence f is undefined on every open interval containing zero. What we can say in the case a = 0 is that √ lim x = 0, x→0+ and that the left-hand limit lim √ x→0− x ◗ does not exist. To each of the limit laws in Section 2.2 there correspond two one-sided limit laws—a right-hand version and a left-hand version. You may apply these one-sided limit laws in the same way you apply the two-sided limit laws in the evaluation of limits. EXAMPLE 7 Figure 2.3.13 shows the graph of the function f defined by ⎧ ⎨x 2 if x 0; 1 f (x) = if x > 0. ⎩x sin x y Clearly lim f (x) = 0 lim f (x) = 0 and x→0− x x→0+ by a one-sided version of the squeeze law (as in Example 4). It therefore follows from Theorem 2 that ◗ lim f (x) = 0. x→0 FIGURE 2.3.13 y = f (x) (Example 7). EXAMPLE 8 Upon applying the appropriate one-sided limit laws, we find that lim x→3− x2 + x2 + 1 9 − x2 lim x 2 = x→3− 2 lim (x + 1) x→3− + lim (9 − x 2 ) x→3− √ 9 9 + 0= . 9+1 10 √ Note that the two-sided limit at 3 is not defined because 9 − x 2 is not defined when x > 3. ◗ = Existence of Tangent Lines Recall that the slope of the line tangent to the graph y = f (x) at the point P(a, f (a)) is defined to be m = lim x→a f (x) − f (a) x −a (10) provided that this (two-sided) limit exists. In this case an equation of the line tangent to the graph y = f (x) at P(a, f (a)) is y − f (a) = m(x − a). If the limit in (10) does not exist, then we say that the curve y = f (x) does not have a tangent line at the point P. The following example gives perhaps the simplest example of a function whose graph has a tangent line everywhere except at a single isolated point. 81 82 CHAPTER 2 Prelude to Calculus EXAMPLE 9 Show that the graph y = |x| has no tangent line at the origin. y Solution Figure 2.3.14 shows the graph of the function f (x) = |x|. The sharp corner at the point (0, 0) makes it intuitively clear that there can be no tangent line there— surely no single straight line can be a good approximation to the shape of the graph at the origin. To verify this intuitive observation, note that when a = 0 we have y = |x| |h| f (a + h) − f (a) −1 = = +1 h h x if h < 0, if h > 0. Hence the left-hand limit of the quotient is −1, whereas the right-hand limit is +1. Therefore the two-sided limit in (10) does not exist, so the graph y = |x| has no tangent line at the origin, where a = 0. ◗ FIGURE 2.3.14 The graph of f (x) = |x| has a corner point at (0, 0). QUESTION Does Fig. 2.3.14 make it clear to you that for f (x) = |x| and a = 0, the value of the “slope limit” in (10) is given by −1 +1 m= if a < 0; if a > 0? It follows (as is apparent from Fig. 2.3.14) that the line y = x is tangent to the graph y = |x| at any point of the graph to the right of the origin, and that the line y = −x is the tangent line at any point of the graph to the left of the origin. Infinite Limits In Example 8 of Section 2.2, we investigated the function f (x) = 1/(x − 1)2 ; the graph of f is shown in Fig. 2.3.15. The value of f (x) increases without bound (that is, eventually exceeds any preassigned number) as x approaches 1 either from the right or from the left. This situation can be described by writing y y= lim 1 (x − 1)2 x→1− 1 1 = +∞ = lim , (x − 1)2 x→1+ (x − 1)2 (11) and we say that each of these one-sided limits is equal to “plus infinity.” x x=1 CAUTION The expression lim x→1+ FIGURE 2.3.15 The graph of the 1 function f (x) = . (x − 1)2 1 = +∞ (x − 1)2 does not mean that there exists an “infinite real number” denoted by +∞—there does not! Neither does it mean that the limit on the left-hand side in Eq. (12) exists—it does not! Instead, Eq. (12) is just a convenient way of saying why the right-hand limit in Eq. (12) does not exist: because the quantity 1/(x − 2)2 increases without bound as x → 1+ . With similar provisos we may write y 1 = +∞ x→1 (x − 1)2 lim y= (12) 1 x x (13) despite the fact that the (two-sided) limit in Eq. (13) does not exist. The expression in Eq. (13) is merely a convenient way of saying that the limit in Eq. (13) does not exist because 1/(x − 1)2 increases without bound as x → 1 from either side. Now consider the function f (x) = 1/x; its graph is shown in Fig. 2.3.16. This function increases without bound as x approaches zero from the right but decreases without bound—it becomes less than any preassigned negative number—as x approaches zero from the left. We therefore write FIGURE 2.3.16 The graph of the 1 function f (x) = . x 82 lim x→0− 1 = −∞ x and lim x→0+ 1 = +∞. x (14) More About Limits SECTION 2.3 83 There is no shorthand for the two-sided limit in this case. We may say only that lim x→0 1 x does not exist. EXAMPLE 10 Investigate the behavior of the function 2x + 1 x −1 near the point x = 1, where the limit of f (x) does not exist. f (x) = Solution First we look at the behavior of f (x) just to the right of the number 1. If x is greater than 1 but close to 1, then 2x + 1 is close to 3 and x − 1 is a small positive number. In this case the quotient (2x + 1)/(x − 1) is a large positive number, and the closer x is to 1, the larger this positive quotient will be. For such x, f (x) increases without bound as x approaches 1 from the right. That is, lim x→1+ 2x + 1 = +∞, x −1 (15) as the data in Fig. 2.3.17 suggest. 10 y= 5 y 2x + 1 x−1 y=2 0 -5 0 x 5 10 2x + 1 x −1 1.1 1.01 1.001 1.0001 .. . ↓ 1 32 302 3002 30002 .. . ↓ +∞ 0.9 0.99 0.999 0.9999 .. . ↓ 1 −28 −298 −2998 −29998 .. . ↓ −∞ 2x + 1 for x x −1 If instead x is less than 1 but still close to 1, then 2x + 1 is still close to 3, but now x − 1 is a negative number close to zero. In this case the quotient (2x + 1)/(x − 1) is a (numerically) large negative number and decreases without bound as x → 1− . Hence we conclude that 2x + 1 = −∞. (16) lim x→1− x − 1 The results in Eqs. (15) and (16) provide a concise description of the behavior of f (x) = (2x + 1)/(x − 1) near the point x = 1. (See Fig. 2.3.18.) Finally, to remain consistent with Theorem 2 on one-sided and two-sided limits, we do not write 2 1.5 2x + 1 = ∞. (Wrong!) x→1 x − 1 Do you see, however, that it would be correct to write 2x + 1 = +∞? lim x→1 x − 1 lim 1 0.5 x near 1 (Example 10). x=1 FIGURE 2.3.18 Graph of 2x + 1 . f (x) = x −1 y 2x + 1 x −1 FIGURE 2.3.17 The behavior of f (x) = -5 -10 -10 x y = log10 x 0 -0.5 -1 -1.5 -2 0 1 2 3 4 5 6 7 8 9 10 x FIGURE 2.3.19 Graph of f (x) = log10 x. ◗ EXAMPLE 11 The graph of f (x) = log10 x is shown in Fig. 2.3.19. The graph makes it clear that lim log10 x = −∞. x→0+ But the left-hand limit of f (x) at x = 0 does not exist because log10 x is not defined if x 0. ◗ 83 84 CHAPTER 2 Prelude to Calculus 5 4 y = 21/x 3 for x > 0 1/x y=2 2 for x < 0 1 y 0 -1 -2 -3 -4 -5 -5 -4 -3 -2 -1 0 1 2 3 4 5 x EXAMPLE 12 Look at the graph of y = 2x in Fig. 1.4.10 to see that lim x→0− 1 = −∞ x lim 21/x = 0 implies that x→0− (because 2t → 0 as t → −∞), whereas lim x→0+ 1 = ∞ implies that x lim 21/x = ∞ x→0+ (because 2t → +∞ as t → +∞). These one-sided limits of 21/x at x = 0 are ◗ illustrated in Fig. 2.3.20. The Basic Trigonometric Limit FIGURE 2.3.20 Graph of f (x) = 21/x . We now provide a geometric proof that sin θ = 1. θ→0 θ (17) lim Proof Figure 2.3.21 shows the angle θ , the triangles OPQ and ORS, and the circular sector OPR that contains the triangle OPQ and is contained in the triangle ORS. Hence y x2 + y2 = 1 area( OPQ) < area(sector OPR) < area( ORS). P θ 1 1 1 sin θ sin θ cos θ < θ < tan θ = . 2 2 2 2 cos θ tanθ sinθ O In terms of θ, this means that S Q cosθ R x FIGURE 2.3.21 Aid to the proof of the basic trigonometric limit. Here we use the standard formula for the area of a triangle to obtain the area of OPQ and ORS. We also use the fact that the area of a circular sector in a circle of radius r is A = 12 r 2 θ if the sector is subtended by a central angle of θ radians; here, r = 1. If 0 < θ < π/2, then we can divide each member of the last inequality by 12 sin θ to obtain 1 θ < . cos θ < sin θ cos θ We take reciprocals, which reverses the inequalities: cos θ < sin θ 1 < . θ cos θ Now we apply the squeeze law of limits with f (θ) = cos θ , g(θ ) = sin θ , θ and h(θ) = 1 . cos θ Because it is clear from Eq. (1) (at the beginning of this section) that f (θ) and h(θ) both approach 1 as θ → 0+ , so does g(θ ) = (sin θ )/θ . This geometric argument shows that (sin θ )/θ → 1 for positive values of θ that approach zero. But the same result follows for negative values of θ, because sin(−θ ) = − sin θ. So we have proved ◆ Eq. (17). The Precise Definition of the Limit When we say that f (x) approaches the limiting value L as x approaches a, we imply that the behavior of the variable x controls the behavior of the value f (x). As x approaches a, this forces the value of f (x) to approach L. In Section 2.2 we said that limx→a f (x) = L provided that we can make f (x) as close to L as we please merely by choosing x sufficiently close to a (though not equal to a). But how close is “sufficiently close”? We can say how close to L we want f (x) to be by prescribing an error tolerance. Then the question is this: How close to a must 84 More About Limits SECTION 2.3 85 x be in order to force the numerical difference | f (x) − L|—the “discrepancy” between f (x) and L—to be less than the prescribed error tolerance. For instance: • • • How close to a must x be to guarantee that | f (x) − L| < 0.1? How close to a must x be to guarantee that | f (x) − L| < 0.01? How close to a must x be to guarantee that | f (x) − L| < 0.001? For any given error tolerance—however small it may be—we need to determine how close to a (but not equal to a) the variable x must be in order to satisfy that error tolerance. EXAMPLE 13 Suppose that a = 2 and f (x) = 5x − 3. We could easily use the limit laws to show that limx→2 (5x − 3) = 7, so that L = 7. But let’s instead begin afresh. We note first that | f (x) − L| = |(5x − 3) − 7| = |5x − 10| = 5 · |x − 2|. Thus |(5x − 3) − 7| is always 5 times |x − 2|. It follows that • • • If |x − 2| < 0.02 then |5x − 10| = 5 · |x − 2| < 5 · (0.02) = 0.1. If |x − 2| < 0.002 then |5x − 10| = 5 · |x − 2| < 5 · (0.002) = 0.01. If |x − 2| < 0.0002 then |5x − 10| = 5 · |x − 2| < 5 · (0.0002) = 0.001. More generally, we need only divide any given error tolerance > 0 by 5 to get the “variance” in x that works: then |(5x − 3) − 7| = 5 · |x − 2| < 5 · = . (18) If |x − 2| < 5 5 Thus we can force f (x) = 5x − 3 to be within of L = 7 merely by requiring that x be within /5 of a = 2. In this example it is also harmless if x = 2 as well—in which case |(5x − 3) − 7| = 0—but we include the requirement that x = 2 by writing 0 < |x − 2| < /5. Finally, if we write δ = /5 for this variance in x that forces an acceptable discrepancy in f (x) = 5x − 3, we conclude from (18) that |(5x − 3) − 7| < for all x such that 0 < |x − 2| < δ. (19) ◗ The exact meaning of limits was debated vigorously—sometimes acrimoniously—during the 17th and 18th centuries. The condition in (19) illustrates the precise definition of the limit that was finally formulated by the German mathematician Karl Weierstrass (1815–1897) and is the definition accepted to this day. DEFINITION The Limit Suppose that f (x) is defined in an open interval containing the point a (except possibly not at a itself). Then we say that the number L is the limit of f (x) as x approaches a—and we write lim f (x) = L x→a —provided that the following criterion is satisfied: Given any number > 0, there exists a corresponding number δ > 0 such that | f (x) − L| < for all x such that 0 < |x − a| < δ. (20) The condition in (20) can be rewritten in the form If 0 < |x − a| < δ then | f (x) − L| < , or even more simply in the form 0 < |x − a| < δ implies that | f (x) − L| < . (21) 85 86 CHAPTER 2 Prelude to Calculus y y = f(x) x=a+δ x=a−δ L+∋ y=L+∋ L L−∋ y=L−∋ a−δ a a+δ x FIGURE 2.3.22 Geometric illustration of the limit definition. Figure 2.3.22 illustrates this definition, which for obvious reasons is often called the “epsilon-delta” definition of limits. The points on the graph of y = f (x) that satisfy the inequality |f (x) − L| < are those that lie between the horizontal lines y = L − and y = L + . The points on this graph that satisfy the inequality |x − a| < δ are those that lie between the vertical lines x = a − δ and x = a + δ. Consequently, the definition of the limit implies that limx→a f (x) = L if and only if the following statement is true: Suppose that the two horizontal lines y = L − and y = L + (with > 0) are given. Then it is possible to choose two vertical lines x = a − δ and x = a + δ (with δ > 0) so that every point (with x = a) on the graph of y = f (x) that lies between the two vertical lines must also lie between the two horizontal lines. Figure 2.3.22 suggests that the closer together are the two horizontal lines, the closer together the two vertical lines will need to be. This is precisely what we mean by “making f (x) closer to L by making x closer to a.” Application of the epsilon-delta definition of limits to establish a limit is usually a two-step process: • • Given > 0, we first analyze the first inequality |f (x) − L| < in (20) to estimate or deduce a value of δ > 0 that works. Then we attempt to prove that this value of δ works—that is, prove that 0 < |x − a| < δ implies that |f (x) − L| < . EXAMPLE 14 Use the epsilon-delta definition of limits to prove that lim (13x − 29) = 10. x→3 Solution Our analysis of the first inequality in (20) consists of noting that it takes the form |(13x − 29) − 10| = |13x − 39| = 13 · |x − 3| < , which boils down to |x − 3| < /13. This leads us to guess—on the basis of rather strong circumstantial evidence—that the value δ = /13 will work. To prove this, we need only note that if δ = /13, then 0 < |x − 3| < δ implies that |(13x − 29) − 10| = 13 · |x − 3| < 13 · Thus 0 < |x − 3| < δ implies that |(13x − 29) − 10| < , as desired. 86 = . 13 ◗ More About Limits SECTION 2.3 EXAMPLE 15 Use the epsilon-delta definition of limits to prove that limx→0 0. √ 3 87 x= Solution Our analysis of the first inequality in (20) consists of noting that it takes the form √ √ 3 x − 0 = 3 x = 3 |x| < , which can be simplified to |x| < 3 . This leads us to guess that the value δ = 3 will work. To prove this, we need only note that if δ = 3 , then √ √ 3 0 < |x − 0| < δ implies that 3 x − 0 = 3 |x| < 3 = . √ ◗ Thus 0 < |x − 0| < δ implies that | 3 x − 0| < , as desired. Given a value of > 0, it is frequently more difficult to guess a value of δ that works than to prove that it does; see Problems 75–84 and this section’s project for additional practice. In Appendix D we use the epsilon-delta definition of limits to establish rigorously the laws of limits. 2.3 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. sin x = 1. x 0 1 − cos x lim = = 1. x→0 x 0 tan 3y tan 3y lim = lim = tan 3. y→0 y→0 y y If x is any real number, then −|x| x |x|. x If f (x) = , then lim f (x) = 1 and lim f (x) = −1. |x| x→0+ x→0− Let g(x) = [[x]] (the greatest integer function). Then lim g(x) does not exist 1. lim x→0 2. 3. 4. 5. 6. 7. 8. 9. 10. x→3 because the left-hand limit of g(x) at x = 3 is not equal to the right-hand limit of g(x) at x = 3. There is no line tangent to the graph of f (x) = |x| at (0, 0). 1 lim = 0. x→0 x 1 lim does not exist. x→0 x 2 1 lim 2 = +∞. x→0 x 2.3 CONCEPTS: QUESTIONS AND DISCUSSION 1. We have interpreted the statement limx→a f (x) = L to mean (roughly) that “ f (x) tends to get closer and closer to L as x gets closer and closer to a.” What would be meant by the statement that “ f (x) gets steadily closer to L as x gets steadily closer to a”? State it precisely, something along the lines that “ f (x) is still closer to L whenever x is still closer to a” (which is still not sufficiently precise). Does this follow from the statement that limx→a f (x) = L? It may help to think about the oscillatory function of Example 4. 2. Formulate precise epsilon-delta definitions of one-sided limits, as well as an Mdelta definition of the infinite limit limx→a f (x) = +∞. The latter definition should involve the inequality f (x) > M; illustrate it with a figure that is similar to Fig. 2.3.22, but involves only a single horizontal line. 87 88 CHAPTER 2 Prelude to Calculus 2.3 PROBLEMS Find the trigonometric limits in Problems 1 through 24. If you have a graphing calculator or a computer with graphing facility, verify that graphical evidence supports your answer. θ θ→0 sin θ 1 − cos θ lim θ→0 θ2 2t lim t→0 (sin t) − t sin 5x lim x→0 x sin x lim √ x→0 x 1. lim 3. 5. 7. 9. x 1 11. lim sin x→0 x 3 1 − cos x 13. lim x→0 sin x 15. lim x sec x csc x x→0 1 − cos θ θ sin θ tan z 19. lim z→0 sin 2z 17. lim θ→0 21. lim x cot 3x x→0 1 t 23. lim 2 sin2 t→0 t 2 sin θ θ→0 θ 2 tan θ lim θ→0 θ sin(2θ 2 ) lim θ→0 θ2 sin 2z lim z→0 z cos 3z 1 − cos 2x lim x→0 x 2 2 2. lim 4. 6. 8. 10. 12. 14. 16. 18. 20. 22. 24. (sin 3θ)2 lim 2 θ→0 θ cos θ tan 3x lim x→0 tan 5x sin 2θ lim θ→0 θ sin2 θ lim θ→0 θ tan 2x lim x→0 3x x − tan x lim x→0 sin x sin 2x lim x→0 sin 5x Use the squeeze law of limits to find the limits in Problems 25 through 28. Also illustrate each of these limits by graphing the functions f , g, and h (in the notation of the squeeze law) on the same screen. 1 25. lim x 2 cos 10x 26. lim x 2 sin 2 x→0 x→0 x √ 1 1 28. lim 3 x sin 27. lim x 2 cos √ 3 x→0 x→0 x x Use one-sided limit laws to find the limits in Problems 29 through 48 or to determine that they do not exist. √ 29. lim (3 − x) 30. lim (4 + 3x 3/2 ) + x→0 x→0+ √ √ 31. lim x − 1 32. lim 4 − x x→1− x→4− 33. lim x2 − 4 34. lim 35. lim x(5 − x) 36. lim x 4 − 4x x −4 38. x→2+ x→5− 37. lim x→4+ x −5 x→5− |x − 5| √ x 2 − 6x + 9 41. lim + x→3 x −3 2−x 43. lim x→2+ |x − 2| 39. lim 88 9 − x2 x→3+ x→2− 40. lim x→−3+ x2 6 − x − x2 16 − x 2 lim √ x→−4+ 16 − x 2 x −2 − 5x + 6 7−x 44. lim x→7− |x − 7| 42. lim x→2+ x2 45. lim x→1+ 47. lim x→5+ 1 − x2 1−x 46. lim x→0− (5 − x)2 5−x 48. x x − |x| 4+x lim (4 + x)2 x→−4− For each of the functions in Problems 49 through 58, there is exactly one point a where both the right-hand and left-hand limits of f (x) fail to exist. Describe (as in Example 10) the behavior of f (x) for x near a. 49. f (x) = 51. f (x) = 53. f (x) = 55. f (x) = 57. f (x) = 1 x −1 x −1 x +1 1 − x2 x +2 |1 − x| (1 − x)2 x −2 4 − x2 50. f (x) = 52. f (x) = 54. f (x) = 56. f (x) = 58. f (x) = 2 3−x 2x − 5 5−x 1 (x − 5)2 x +1 x 2 + 6x + 9 x −1 x 2 − 3x + 2 In Problems 59 and 60, find the left-hand and right-hand limits of f (x) at a = 2. Does the two-sided limit of f exist there? Sketch the graph of y = f (x). x2 − 4 |x − 2| x 4 − 8x + 16 60. f (x) = |x − 2| 59. f (x) = Problems 61 through 68, do the following: (a) Sketch the graph of the given function f . (b) For each integer n, evaluate the one-sided limits lim f (x) x→n − and lim f (x) x→n + in terms of n. (c) Determine those values of a for which lim f (x) exists. x→a Recall that [[x]] denote the greatest integer that does not exceed x. 61. f (x) = 2 2 + (−1)x 62. f (x) = x 0 if x is not an integer; if x is an integer. if x is not an integer; if x is an integer. 63. f (x) = [[10x]] 64. f (x) = (−1)[[x]] 65. f (x) = x − [[x]] − 1 2 x 2 67. f (x) = [[x]] + [[−x]] ⎧ ⎨ [[x]] if x = 0; 68. f (x) = x ⎩ 0 if x = 0. 66. f (x) = 1 [[10x]], the value of x to one decimal place 10 rounded down, sketch the graph of g and determine the values of a such that lim g(x) exists. 69. If g(x) = x→a More About Limits SECTION 2.3 89 76. lim (17x − 35) = 50 x→5 √ 77. lim x = 0 Suggestion: First formulate a precise 70. The sign function sgn(x) is defined as follows: x if x = 0; sgn(x) = |x| 0 if x = 0. x→0+ epsilon-delta definition of right-hand limits. Use the sign function to define two functions f and g whose limits as x → 0 do not exist, but such that (a) lim [ f (x) + g(x)] does exist; 78. lim x 2 = 0 x→0 79. lim x 2 = 4 x→2 x→0 (b) lim f (x) · g(x) 71. Let |x 2 − 4| = |x + 2| · |x − 2|. does exist. x→0 x2 f (x) = 0 Then argue that if we agree to choose δ < 1, then |x −2| < δ will imply that |x + 2| < 5. (Why?) Then show that it works to choose δ to be the smaller of the two numbers 1 and /5. if x is rational; if x is irrational. Use the squeeze law of limits to show that lim f (x) = x→0 f (0) = 0. 72. Sketch the graph of the function f (x) = 80. lim (x 2 − 5x − 4) = 10 x→7 Then argue that if we agree to choose δ < 1, then |x −7| < δ will imply that |x + 2| < 10. (Why?) 1 1 + 21/x 81. lim (2x 2 − 13x − 25) = 45 x→10 x→0 In Problems 73 and 74, first examine the value of f (x) on intervals of the form 1 1 <x< n+1 n where n is an integer. Then determine whether or not lim f (x) In Problems 75 through 84, use the epsilon-delta definition of limits to prove the given equation. Suggestion: Write |(2x 2 − 13x − 25) − 45| = |2x + 7| · |x − 10|. Then argue that if we agree to choose δ < 1, then |x − 10| < δ will imply that |2x + 7| < 29. (Why?) 82. lim x 3 = 8 x→2 x→0 exists. If your graphing calculator or computer has a greatest integer (or “floor”) function, graph f to corroborate your answer. 1 1 2 73. f (x) = x · 74. f (x) = x · x x x→−3 Suggestion: Note that |(x 2 − 5x − 4) − 10| = |x + 2| · |x − 7|. for x = 0. Then determine whether or not lim f (x) exists. 75. lim (7x − 9) = −30 Suggestion: Note that Suggestion: First verify that |x 3 − 8| = |x 2 + 2x + 4| · |x − 2|. Then argue that if we agree to choose δ < 1, then |x −2| < δ will imply that |x 2 + 2x + 4| < 19. (Why?) 83. Generalize the approach of Problem 79 to prove that lim x 2 = a 2 . x→a 84. Generalize the approach of Problem 82 to prove that lim x 3 = a 3 . x→a 2.3 INVESTIGATION: Numerical Epsilon-Delta Limits Figure 2.3.23 shows a steadily rising graph y = f (x) that passes through the point (a, L). Given a single numerical value of > 0, we can illustrate the limit limx→a f (x) = L by solving the equations f (x) = L ± graphically or numerically for the indicated values x1 to the left of a such that f (x1 ) = L − and x2 to the right of a such that f (x2 ) = L + . If δ > 0 is chosen smaller than either of the two indicated distances δ1 = a − x1 and δ2 = x2 − a, then the figure suggests that 0 < |x − a| < δ implies that | f (x) − L| < . (21) You should understand that an actual proof that limx→a f (x) = L must show that, given any > 0 whatsoever, there exists a δ > 0 that works for this —meaning that the implication in (21) holds. Doing it for a single value of does not constitute a proof, but doing it for several successively smaller values of can be instructive and perhaps convincing. Suppose, for example, that f (x) = x 3 + 5x 2 + 10x + 98, a = 3, and L = 200. Then, for a particular fixed value of > 0, you can use a calculator or computer 89 90 CHAPTER 2 Prelude to Calculus y y = f(x) L+∋ ∋ (a, L) L ∋ L−∋ δ1 δ2 x1 = a − δ 1 x a x2 = a + δ 2 FIGURE 2.3.23 Finding δ = min(δ1 , δ2 ) graphically. algebra system to solve the equations x 3 + 5x 2 + 10x + 98 = 200 − and x 3 + 5x 2 + 10x + 98 = 200 + numerically for the solutions x1 and x2 near 3. With = 1, = 0.2, and = 0.04 you should obtain the following results. x1 x2 δ1 δ2 δ 1 0.2 0.04 2.98503 2.99701 2.99940 3.01488 3.00298 3.00060 0.01497 0.00299 0.00060 0.01488 0.00298 0.00060 0.01 0.002 0.0005 In the final column, each value of δ is (for safety) chosen a bit smaller than either δ1 or δ2 , to be sure that it works with the corresponding value of . You might try a still smaller value such as = 0.001 to find a corresponding value of δ that works. Then carry out a similar investigation to “verify ” numerically a polynomial limit of your own selection. 2.4 THE CONCEPT OF CONTINUITY Anyone can see a drastic difference between the graphs in Figs. 2.4.1 and 2.4.2. Figure 2.4.1 is intended to suggest that the graph y = f (x) can be traced with a continuous motion—without any jumps—of the pen from left to right. But in Fig. 2.4.2 the pen must make a sudden jump at x = a. The concept of continuity isolates the property that the function f of Fig. 2.4.1 possesses but that the function g of Fig. 2.4.2 lacks. We first define continuity of a function at a single point. y y y = f(x) y = g(x) x a FIGURE 2.4.1 A continuous graph. 90 FIGURE 2.4.2 A graph that is not continuous. x The Concept of Continuity SECTION 2.4 91 DEFINITION Continuity at a Point Suppose that the function f is defined in a neighborhood of a. We say that f is continuous at a provided that limx→a f (x) exists and, moreover, that the value of this limit is f (a). In other words, f is continuous at a provided that lim f (x) = f (a). x→a (1) Briefly, continuity of f at a means this: The limit of f at a is equal to the value of f there. Another way to put it is this: The limit of f at a is the “expected” value—the value that you would assign if you knew the values of f for x = a in a neighborhood of a and you knew f to be “predictable.” Alternatively, continuity of f at a means this: When x is close to a, f (x) is close to f (a). Analysis of the definition of continuity shows us that to be continuous at the point a, the function f must satisfy the following three conditions: 4 1. The function f must be defined at a [so that f (a) exists]. 2. The limit of f (x) as x approaches a must exist. 3. The numbers in conditions 1 and 2 must be equal: y 0 f(x) = 1/(x − 2) lim f (x) = f (a). -4 x→a -4 0 x If any one of these conditions is not satisfied, then f is not continuous at a. Examples 1 through 3 illustrate these three possibilities for discontinuity at a point. If the function f is not continuous at a, then we say that it is discontinuous there, or that a is a discontinuity of f . Intuitively, a discontinuity of f is a point where the graph of f has a “gap,” or “jump,” of some sort. 4 FIGURE 2.4.3 The function f (x) = 1/(x − 2) has an infinite discontinuity at x = 2 (Example 1). EXAMPLE 1 Figure 2.4.3 shows the graph of the function f defined by y 1 for x = 2. x −2 Because f is not defined at the point x = 2, it is not continuous there. Moreover, f ◗ has what might be called an infinite discontinuity at x = 2. f (x) = (0, 1) x (0, −1) (not on the graph) EXAMPLE 2 Figure 2.4.4 shows the graph of the function g defined by FIGURE 2.4.4 The function g has a finite jump discontinuity at x = 0 (Example 2). 2 +1 if x 0; −1 if x < 0. Its left-hand and right-hand limits at x = 0 are unequal, so g(x) has no limit as x → 0. Consequently, the function g is not continuous at x = 0; it has what might be called a ◗ finite jump discontinuity there. EXAMPLE 3 Figure 2.4.5 shows the graph of the function h defined by ⎧ ⎨ sin x if x = 0; h(x) = x ⎩ 0 if x = 0. 1 y = h(x) y 0 g(x) = sgn(x) = (0, 0) -1 Because we saw in Section 2.3 that -2 -10 -5 0 x 5 10 FIGURE 2.4.5 The point (0, 0) is on the graph; the point (0, 1) is not (Example 3). sin x = 1, x whereas h(0) = 0, we see that the limit and the value of h at x = 0 are not equal. Thus the function h is not continuous there. As x moves from negative values through x = 0 to positive values, the value of h(x) jumps from “near 1” to zero and back again. ◗ lim h(x) = lim x→0 x→0 91 92 CHAPTER 2 Prelude to Calculus The discontinuity at the origin in Example 3 is an example of a removable discontinuity. The point a where the function f is discontinuous is called a removable discontinuity provided that there exists a function F such that • • F(x) = f (x) for all x = a in the domain of definition of f , and This new function F is continuous at a. The original function f may or may not be defined at a, but in any event the graphs of f and F differ only at x = a. Sometimes it is simpler to speak of “old” and “new” versions of the same function f . Thus we might say that a removable discontinuity of a function is one that can be removed by suitable definition—or, if necessary, redefinition—of the function at that single point. REMARK The discontinuity at the origin of the function h in Example 3 is removable. The reason is that if we change the original value h(0) = 0 to h(0) = 1, then sin x = 1 = h(0), h→0 x lim h(x) = lim x→0 so h is now continuous at x = 0. By contrast, the discontinuities in the sawtooth function f of the next example are not removable, because we see genuine jumps or gaps in the graph that obviously cannot be removed simply by changing the values of f at these discontinuities. EXAMPLE 4 Figure 2.4.6 shows the graph of the function f defined by y f (x) = x − [[x]]. x FIGURE 2.4.6 The “sawtooth function” of Example 4. As before, [[x]] denotes the largest integer no greater than x. If x = n, an integer, then [[n]] = n, so f (n) = 0. On the open interval (n, n + 1), the graph of f is linear and has slope 1. It should be clear that f is • • Continuous at x if x is not an integer; Discontinuous at each integer point on the x-axis. ◗ Combinations of Continuous Functions Frequently we are most interested in functions that are continuous. Suppose that the function f is defined on an open interval or a union of open intervals. Then we say simply that f is continuous if it is continuous at each point of its domain of definition. It follows readily from the limit laws in Section 2.2 that any constant multiple, sum, difference, or product of continuous functions is continuous. That is, if c is a constant and the functions f and g are continuous at a, then so are the functions c f, f + g, f − g, and f · g. For instance, if f and g are continuous at a, then lim [ f (x) + g(x)] = lim f (x) + lim g(x) = f (a) + g(a), x→a x→a x→a so it follows that the sum f + g is also continuous at a. EXAMPLE 5 Because f (x) = x and constant-valued functions are clearly continuous everywhere, it follows that the cubic polynomial function f (x) = x 3 − 3x 2 + 1 = x · x · x + (−3) · x · x + 1 is continuous everywhere. More generally, it follows in a similar way that every polynomial function p(x) = bn x n + bn−1 x n−1 + · · · + b1 x + b0 92 ◗ The Concept of Continuity SECTION 2.4 93 is continuous at each point of the real line. In short, every polynomial is continuous everywhere. If p(x) and q(x) are polynomials, then the quotient law for limits and the continuity of polynomials imply that lim x→a lim p(x) p(a) p(x) x→a = = q(x) lim q(x) q(a) x→a 4 y 0 provided that q(a) = 0. Thus every rational function y = 1/(x − 2)2 f (x) = -4 -4 0 x 4 FIGURE 2.4.7 The function f (x) = 1/(x − 2)2 has an infinite discontinuity at x = 2. p(x) q(x) (2) is continuous wherever it is defined—that is, wherever the denominator polynomial is nonzero. More generally, the quotient of any two continuous functions is continuous at every point where the denominator is nonzero. At a point x = a where the denominator in Eq. (2) is zero, q(a) = 0, there are two possibilities: • • If p(a) = 0, then f has an infinite discontinuity (as in Figs. 2.4.3 and 2.4.7) at x = a. Otherwise, f may have a removable discontinuity at x = a. EXAMPLE 6 Suppose that f (x) = y = F(x) 4 (2, 1) y 0 x2 x −2 . − 3x + 2 (3) We factor the denominator: x 2 − 3x + 2 = (x − 1)(x − 2). This shows that f is not defined at x = 1 and at x = 2. Thus the rational function defined in Eq. (3) is continuous except at these two points. Because cancellation gives f (x) = x2 1 x −2 = − 3x + 2 x −1 except at the single point x = 2, the new function -4 -4 0 x 4 FIGURE 2.4.8 In Example 6, the graph y = F(x) consists of the graph y = f (x) with the single point (2, 1) adjoined. F(x) = 1 x −1 (4) agrees with f (x) if x = 2 but is continuous at x = 2 also, where F(2) = 1. Thus f has a removable discontinuity at x = 2; the discontinuity at x = 1 is not removable. ◗ (See Fig. 2.4.8.) Continuity of Trigonometric Functions At the beginning of Section 2.3 we noted that lim cos x = 1 x→0 and lim sin x = 0. x→0 (5) Because cos 0 = 1 and sin 0 = 0, the sine and cosine functions are continuous at x = 0 by definition. But this fact implies that they are continuous everywhere. THEOREM 1 Continuity of Sine and Cosine The functions f (x) = sin x and g(x) = cos x are continuous functions of x on the whole real line. 93 94 CHAPTER 2 Prelude to Calculus Proof We give the proof only for sin x; the proof for cos x is similar. (See Problem 67.) We want to show that limx→a sin x = sin a for every real number a. If we write x = a + h, so that h = x − a, then h → 0 as x → a. Thus we need only show that lim sin(a + h) = sin a. h→0 But the addition formula for the sine function yields lim sin(a + h) = lim (sin a cos h + cos a sin h) h→0 h→0 = (sin a) lim cos h + (cos a) lim sin h h→0 h→0 = sin a ◆ as desired; we used the limits in Eq. (5) in the last step. REMARK It now follows that the function tan x = sin x cos x (6) is continuous except where cos x = 0—that is, except when x is an odd integral multiple of π/2. As illustrated in Fig. 2.4.9, tan x has an infinite discontinuity at each such point. 4 y= = tan x 2 y 0 − π 2 π 2 3π 2 -2 -4 -4 -2 0 2 4 6 x FIGURE 2.4.9 The function tan x has infinite discontinuities at x = ±π/2, ±3π/2, . . . . Composition of Continuous Functions Recall from Section 1.4 that the composition of the two functions f and g is the function h = f ◦ g defined by h(x) = f (g(x)) for all x in the domain of g such that u = g(x) is in the domain of f . Theorem 2 implies that functions built by forming compositions of continuous functions are themselves continuous. THEOREM 2 Continuity of Compositions The composition of two continuous functions is continuous. More precisely, if g is continuous at a and f is continuous at g(a), then f ◦ g is continuous at a. 94 The Concept of Continuity SECTION 2.4 95 Proof The continuity of g at a means that g(x) → g(a) as x → a, and the continuity of f at g(a) implies that f (x) → f (g(a)) as x → g(a). Hence the substitution law for limits (Section 2.2) yields lim f (g(x)) = f lim g(x) = f (g(a)), x→a x→a ◆ as desired. Recall from the root law in Section 2.2 that √ √ lim n x = n a x→a under the conditions √ that n is an integer and that a > 0 if n is even. Thus the nth-root function f (x) = n x is continuous everywhere if n is odd; f is continuous for x > 0 if n is even. We may combine this result with Theorem 2. Then we see that a root of a continuous function is continuous wherever it is defined. That is, the composition h(x) = √ n g(x) = [g(x)]1/n √ of f (x) = n x and the √ function g(x) is continuous at a if g is, assuming that g(a) > 0 if n is even (so that n g(a) is defined). EXAMPLE 7 Show that the function f (x) = x −7 x 2 + 2x + 2 2/3 is continuous on the whole real line. Solution Note first that the denominator 5 x 2 + 2x + 2 = (x + 1)2 + 1 (−1, 4) 4 3 y2 1 0 −10 (7, 0) −5 5 0 x FIGURE 2.4.10 The graph 2/3 x −7 . y= x 2 + 2x + 2 10 15 is never zero, because its smallest value (when x = −1) is 02 + 1 = 1. Hence the rational function x −7 r (x) = 2 x + 2x + 2 is defined and continuous everywhere. It then follows from Theorem 2 and the continuity of the cube root function that 3 f (x) = [r (x)]2/3 = [r (x)]2 is continuous everywhere—as suggested by its graph in Fig. 2.4.10, where we see a high point apparently near the point (−1, 4) and the single point (7, 0) where the curve ◗ touches the x-axis. EXAMPLE 8 (a) The exponential function f (x) = 2x is continuous everywhere, and therefore so is the composition h(x) = 2sin x of f and the sine function. Refer to Fig. 2.4.11, where we see high and low points on the graph of y = 2sin x corresponding to the high and low points on the graph of y = sin x. (b) By contrast, the tangent function tan x has infinite discontinuities at odd integral multiples of π/2 (as shown in Fig. 2.4.9), and we see corresponding discontinuities in the composition h(x) = 2tan x when we look at the graph in Fig. 2.4.12. These discontinuities are interesting in that, if a is an odd integral multiple of π/2, then lim h(x) = lim 2tan x = +∞, x→a − x→a − whereas lim h(x) = lim 2tan x = 0. x→a + x→a + ◗ 95 96 CHAPTER 2 Prelude to Calculus 10 3 y = 2tan x y = 2sin x 2 5 y y 1 0 0 -1 -3π/2 -π/2 0 π/2 x 3π/2 -5 -3π/2 5π/2 FIGURE 2.4.11 The function h(x) = 2sin x is continuous everywhere. -π/2 0 π/2 x 3π/2 5π/2 FIGURE 2.4.12 The function h(x) = 2tan x has infinite discontinuities. The function h(x) = 2tan x of Example 8(b) illustrates the concept of one-sided continuity. It is convenient to say that the function f is • continuous from the left at a if lim = f (a), and is • continuous from the right at a if lim = f (a). x→a − x→a + Suppose we define the “augmented function” H by H (x) = 2tan x unless x is an odd integral multiple a of π/2, in which case H (a) = 0. Then it follows from Example 8(b) that H is continuous from the right at a, but is not continuous from the left at a. Of course, a function is automatically continuous at a point if it is continuous from both sides there. √ REMARK We have observed that the function√f (x) = x is continuous for x >√0. However, f is not continuous at x = 0 because x is not defined for x < 0, so lim x x→0 √ √ does not exist. However, lim x = 0 = 0, so the function f is continuous from the x→0+ √ right at 0. Thus x is continuous from the right where it is only defined on the right. √ Hence it is sometimes said—by a slight “abuse of terminology”—that the function x is continuous wherever it is defined. Continuous Functions on Closed Intervals An applied problem typically involves a function whose domain is a closed interval. For example, in the animal pen problem of Section 1.1, we found that the area A of the rectangular pen in Fig. 2.4.13 was expressed as a function of its base length x by x $5/ft y $5/ft $5/ft y $1/ft x Wall FIGURE 2.4.13 The animal pen. A = f (x) = 35 x(30 − x). Although this formula for f is meaningful for all x, only values in the closed interval [0, 30] correspond to actual rectangles, so only such values are pertinent to the animal pen problem. The function f defined on the closed interval [a, b] is said to be continuous on [a, b] provided that • • • f is continuous at each point of the open interval (a, b), f is continuous from the right at the left-hand endpoint a, and f is continuous from the left at the right-hand endpoint b. The last two conditions imply that, at each endpoint, the value of the function is equal to its limit from within the interval. For instance, every polynomial is continuous on √ x is continuous from the every closed interval. The square root function f (x) = √ √ right at 0 because limx→0+ x = 0 = 0 . Therefore f is continuous on the closed interval [0, 1] even though f is not defined for x < 0. 96 The Concept of Continuity SECTION 2.4 97 Continuous functions defined on closed intervals have very special properties. For example, every such function has the intermediate value property of Theorem 3. (A proof of this theorem is given in Appendix E.) We suggested earlier that continuity of a function is related to the possibility of tracing its graph without lifting the pen from the paper. Theorem 3, the intermediate value theorem, expresses this fact with precision. THEOREM 3 Intermediate Value Property Suppose that the function f is continuous on the closed interval [a, b]. Then f (x) assumes every intermediate value between f (a) and f (b). That is, if K is any number between f (a) and f (b), then there exists at least one number c in (a, b) such that f (c) = K . Figure 2.4.14 shows the graph of a typical continuous function f whose domain is the closed interval [a, b]. The number K is located on the y-axis, somewhere between f (a) and f (b). In the figure f (a) < f (b), but this is not important. The horizontal line through K must cross the graph of f somewhere, and the x-coordinate of the point where graph and line meet yields the value of c. The number c is the one whose existence is guaranteed by the intermediate value property of the continuous function f . y y = f(x) f (b) y=K f (a) a c b FIGURE 2.4.14 The continuous function f attains the intermediate value K at x = c. y (1, 1) y = 12 (−1, 0) x Thus the intermediate value theorem implies that each horizontal line meeting the y-axis between f (a) and f (b) must cross the graph of the continuous function f somewhere. This is a way of saying that the graph has no gaps or jumps, suggesting that the idea of being able to trace such a graph without lifting the pen from the paper is accurate. EXAMPLE 9 The discontinuous function defined on [−1, 1] as FIGURE 2.4.15 This discontinuous function does not have the intermediate value property (Example 9). f (x) = 0 1 if x < 0, if x 0 does not attain the intermediate value 12 . See Fig. 2.4.15. ◗ Existence of Solutions of Equations An important application of the intermediate value theorem is the verification of the existence of solutions of equations written in the form f (x) = 0. (7) √ EXAMPLE 10 You could attempt to approximate the number 2 graphically by zooming in on the intersection of the parabola y = x 2 − 2 with the positive x-axis 97 98 CHAPTER 2 Prelude to Calculus (Fig. 2.4.16). The x-coordinate of the intersection yields the positive solution of the equation (2, 2) 2 f (x) = x 2 − 2 = 0. y = x2 − 2 (8) Perhaps it makes no sense to zoom in on this point unless we know that it’s “really there.” But we can see from Eq. (8) that y 0 f (1) = −1 < 0, (1, −1) 0 x f (2) = 2 > 0. We note that the function f is continuous on [1, 2] (it is continuous everywhere) and that K = 0 is an intermediate value of f on the interval [1, 2]. Therefore, it follows from Theorem 3 that f (c) = c2 − 2 = 0 for some number c in (1, 2)—that is, that -2 -2 whereas 2 c2 = 2. FIGURE 2.4.16 The graph of f (x) = x 2 − 2 (Example 10). This number c is the desired square root of 2. Thus it is the intermediate √ value property of continuous functions that guarantees the existence of the number 2: There is a real ◗ number whose square is 2. As indicated in Fig. 2.4.17, the solutions of Eq. (7) are simply the points where the graph y = f (x) crosses the x-axis. Suppose that f is continuous and that we can find a closed interval [a, b] (such as the interval [1, 2] of Example 10) such that the value of f is positive at one endpoint of [a, b] and negative at the other. That is, suppose that f (x) changes sign on the closed interval [a, b]. Then the intermediate value property ensures that f (x) = 0 at some point of [a, b]. y y = f(x) x Solution EXAMPLE 11 The graph y = x 3 − x − 2 shown in Fig. 2.4.18 indicates that the equation f (x) = x 3 − x − 2 = 0 FIGURE 2.4.17 The solution of the equation f (x) = 0. 10 8 6 4 2 y 0 -2 -4 -6 -8 -10 -3 has a solution somewhere between x = 1 and x = 2. Apply the intermediate value theorem to show that this actually is so. Solution The function f (x) is continuous on [1, 2] because it is a polynomial and, therefore, is continuous everywhere. Because f (1) = −2 and f (2) = +4, the intermediate value theorem implies that every number between −2 and +4 is a value of f (x) on [1, 2]. In particular, (2, 4) −2 = f (1) < 0 < f (2) = +4, (1, −2) so the intermediate value property of f implies that f attains the value 0 at some number c between x = 1 and x = 2. That is, y = x3 − x − 2 -2 -1 0 x f (c) = c3 − c − 2 = 0, 1 2 FIGURE 2.4.18 The equation x 3 − x − 2 = 0 of Example 11 appears to have a solution somewhere between x = 1 and x = 2. 3 so x = c is a solution in (1, 2) of the equation x 3 − x − 2 = 0. ◗ The following example shows that not every suspected root of an equation f (x) = 0 that seems to be visible on a computer-plotted figure is actually there. Indeed, a graphing calculator or computer ordinarily is programmed to plot close but isolated points on the desired graph y = f (x) and then join these points with line segments so short that the result looks like a smooth curve. In effect, the computer is assuming that the function f is continuous, whether or not it actually is continuous. EXAMPLE 12 Figure 2.4.19 shows a computer plot of the graph of the function f (x) = 10 · [[1000x]] − 4995 . 10000 The graph y = f (x) appears indistinguishable from the line y = x − 12 , and in particular it appears that the equation f (x) = 0 has the solution x = 12 . But when we zoom in near this alleged solution we see the graph shown in Fig. 2.4.20. Now we see 98 The Concept of Continuity SECTION 2.4 99 0.01 0.008 0.006 0.004 0.002 y 0 -0.002 -0.004 -0.006 -0.008 -0.01 0.49 1 0.5 y ? 0 -0.5 -1 y = f(x) -0.5 0 x 0.5 1 FIGURE 2.4.19 The graph y = f (x) of Example 12 appears to have x-intercept x = 0.5. 0.495 0.5 x 0.505 0.51 FIGURE 2.4.20 The graph in Example 12 jumps across the x-axis—there is no x-intercept. that the function f is discontinuous, and actually “jumps” across the x-axis without ◗ intersecting it. Thus the equation f (x) = 0 has no solution at all. 2.4 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If the limit of the function f at x = a exists and is equal to f (a), then f is continuous at x = a. 1 , then f is not continuous at x = 2. 2. If f (x) = x −2 ⎧ ⎨ sin x if x = 0, then h is continuous at x = 0. 3. If h(x) = x ⎩1 if x = 0, 4. Every polynomial function is continuous at every real number. 5. Every rational function is continuous wherever it is defined. 6. If f is continuous at x = a, then lim f (x) = f (a). x→a 7. The sine and cosine functions are continuous on the entire real line. 8. The composition f ◦ g of the continuous functions f and g is continuous. 9. The function f is said to be continuous on the closed and bounded interval [a, b] provided that f is continuous on (a, b) and, morever, lim f (x) = f (a) x→a + and lim f (x) = f (b). x→b− 10. If f is continuous on the interval [a, b] and K is between f (a) and f (b), then K = f (c) for some number c in (a, b). 2.4 CONCEPTS: QUESTIONS AND DISCUSSION 1. Suppose that a < b < c. If the function f is continuous both on the closed interval [a, b] and on the closed interval [b, c], does it follow that f is continuous on [a, c]? If f is continuous on the closed interval [n, n + 1] for every integer n, does it follow that f is continuous on the entire real line? 2. Suppose that the function f is continuous everywhere and that the composition f (g(x)) is continuous at x = a. Does it follow that g(x) is continuous at a? Suggestion: Consider the possibility that f (x) = |x|. 3. Suppose that p(x) is a polynomial of odd degree with positive leading coefficient. Then its graph y = p(x) “heads southwest in the third quadrant” and 99 100 CHAPTER 2 Prelude to Calculus “northeast in the first quadrant.” Can you state more precisely what these intuitive statements mean? Why does it follow that the equation p(x) = 0 always has at least one solution? 4. If 10 L = y, then we call the number L the base 10 logarithm of y and write L = log y. Assume that the exponential function f (x) = 10x is continuous everywhere (it is!) and—as suggested by its graph in Fig. 1.4.10—it has both arbitrarily large positive values and values arbitrarily close to zero. Then explain why the intermediate value theorem implies that every positive number has a base 10 logarithm. 2.4 PROBLEMS In Problems 1 through 8, apply the limit laws and the theorems of this section to show that the given function is continuous for all x. 1. f (x) = 2x 5 − 7x 2 + 13 2. f (x) = 7x 3 − (2x + 1)5 2x − 1 3. g(x) = 2 4x + 1 √ 5. h(x) = x 2 + 4x + 5 1 − sin x 7. f (x) = 1 + cos2 x x3 4. g(x) = 2 x + 2x + 5 √ 6. h(x) = 3 1 − 5x 8. g(x) = 4 1 − sin2 x In Problems 9 through 14, apply the limit laws and the theorems of this section to show that the given function is continuous on the indicated interval. 1 9. f (x) = , x > −1 x +1 x −1 10. f (x) = 2 , −2 < x < 2 x −4 √ 11. g(t) = 9 − 4t 2 , − 32 t 32 √ 12. h(z) = (z − 1)(3 − z), 1 z 3 x , − 12 π < x < 12 π 13. f (x) = cos x √ 14. g(t) = 1 − 2 sin t, − 16 π < t < 16 π In Problems 15 through 36, tell where the given function is continuous. Recall that when the domain of a function is not specified, it is the set of all real numbers for which the formula of the function is meaningful. 15. f (x) = 2x + √ 3 x 1 x +3 1 19. f (x) = 2 x +1 x −5 21. f (x) = |x − 5| 17. f (x) = x2 + 4 23. f (x) = x −2 x +1 25. f (x) = 3 x −1 3 27. f (x) = 2 x −x 29. f (x) = √ 100 x 4 − x2 16. g(x) = x 2 + 1 x 5 5−t 1 20. g(z) = 2 z −1 x2 + x + 1 22. h(x) = x2 + 1 √ 24. f (t) = 4 4 + t 4 18. f (t) = 26. F(u) = √ 3 3 − u3 √ 28. f (z) = 9 − z 2 1 − x2 30. f (x) = 4 − x2 sin x x2 1 33. f (x) = sin 2x θ cos θ √ 34. f (x) = sin x 35. f (x) = sin |x| 1 36. G(u) = √ 1 + cos u 31. f (x) = 32. g(θ) = In Problems 37 through 48, find the points where the given function is not defined and is therefore not continuous. For each such point a, tell whether or not this discontinuity is removable. x t 38. f (t) = 2 (x + 3)3 t −1 u+1 x −2 40. G(u) = 2 f (x) = 2 x −4 u −u−6 |x − 1| 1 f (x) = 42. h(x) = 1 − |x| (x − 1)3 x − 17 f (x) = |x − 17| x 2 + 5x + 6 g(x) = x +2 −x if x < 0; f (x) = 2 if x > 0 x 37. f (x) = 39. 41. 43. 44. 45. x + 1 if x < 1; 3 − x if x > 1 ⎧ ⎨1 + x 2 if x < 0; 47. f (x) = sin x ⎩ if x > 0 x 1 − cos x if x < 0; 48. f (x) = x if x > 0 x2 46. f (x) = In Problems 49 through 52, find a value of the constant c so that the function f (x) is continuous for all x. 49. f (x) = x +c 4 − x2 if x < 0, if x 0 50. f (x) = 2x + c 2c − x if x 3, if x > 3 51. f (x) = c2 − x 2 2(x − c)2 52. f (x) = c3 − x 3 c sin x if x < 0, if x 0 if x π, if x > π In Problems 53 through 58, apply the intermediate value property of continuous functions to show that the given equation has a solution in the given interval. The Concept of Continuity SECTION 2.4 101 71. Figure 2.4.21 suggests that the equation x = cos x has a solution in the interval (0, π/2). Use the intermediate value theorem to show that this is true. Then use your calculator to approximate this solution accurate to two decimal places. x 2 − 5 = 0 on [2, 3] x 3 + x + 1 = 0 on [−1, 0] x 3 − 3x 2 + 1 = 0 on [0, 1] x 3 = 5 on [1, 2] x 4 + 2x − 1 = 0 on [0, 1] x 5 − 5x 3 + 3 = 0 on [−3, −2] 3 2 In Problems 59 and 60, show that the given equation has three distinct roots by calculating the values of the left-hand side at x = −3, −2, −1, 0, 1, 2, and 3 and then applying the intermediate value property of continuous functions on appropriate closed intervals. 59. x 3 − 4x + 1 = 0 60. x 3 − 3x 2 + 1 = 0 61. Suppose that you accept a job now (time t = 0) at an annual salary of $25, 000 and are promised a 6% raise at the end of each year of employment. Explain why your salary in thousands of dollars after t years is given by the formula [[t]] S(t) = 25 · (1.06) . 62. 63. 64. 65. 66. 67. 68. 69. 70. Graph this function for the first 5 years and comment on its continuity. Suppose that you accept the same job as in Problem 61, but now are promised a 1.5% raise at the end of each quarter (three months). (a) Write a formula for your salary (in thousands of dollars) after t years. (b) Graph this new salary function and comment on its continuity. (c) Which is the better deal, the promised salary of Problem 61 or the one of this problem? Suppose that f and g are two functions both continuous on the interval [a, b], and such that f (a) = g(b) = p and f (b) = g(a) = q where p = q. Sketch typical graphs of two such functions. Then apply the intermediate value theorem to the function h(x) = f (x) − g(x) to show that f (c) = g(c) at some point c of (a, b). Suppose that today you leave your home in Estes Park, CO at 1 P. M . and drive to Grand Lake, arriving at 2 P. M . Tomorrow you leave your destination in Grand Lake at 1 P. M . and retrace the same route, arriving home at 2 P. M . Use Problem 63 as a suggestion to show that at some instant between 1 and 2 P. M . you are at precisely the same point on the road both days. What must you assume about the functions describing your location as a function of time each day? Apply the intermediate value property of continuous functions to show that every positive number a has a square root. That is, given a > 0, prove that there exists a number r such that r 2 = a. Apply the intermediate value property to prove that every real number has a cube root. Show that the cosine function is continuous on the set of all real numbers. (Suggestion: Alter the proof of Theorem 1 of the continuity of the sine function.) Determine where the function f (x) = x +[[x]] is continuous. Suppose that f (x) = 0 if x is a rational number, whereas f (x) = 1 if x is irrational. Prove that f is discontinuous at every real number. Suppose that f (x) = 0 if x is a rational number, whereas f (x) = x 2 if x is irrational. Prove that f is continuous only at the single point x = 0. y=x 1 y 53. 54. 55. 56. 57. 58. y = cos x 0 -1 -2 -3 -3 -2 -1 0 x 1 2 3 FIGURE 2.4.21 The graphs y = x and y = cos x (Problem 71). 72. Figure 2.4.22 suggests that the equation x = −5 cos x has at least three distinct solutions. Use the intermediate value theorem to show that this is true. Then use your calculator to approximate each of these solutions accurate to two decimal places. 8 6 y=x 4 2 y 0 -2 -4 -6 y = −5 cos x -8 -8 -6 -4 -2 0 2 4 6 8 x FIGURE 2.4.22 The graphs y = x and y = −5 cos x (Problem 72). Investigate the continuity of each of the functions defined in Problems 73 through 78. For each discontinuity, determine whether the given function is continuous from the right and whether it is continuous from the left. Use a graphing calculator or computer if you find it helpful. 73. f (x) = 21/x if x = 0; f (0) = 0 74. f (x) = 2−1/x if x = 0; f (0) = 0 1 if x = 0; f (0) = 1 75. f (x) = 1 + 21/x 1 76. f (x) = if x = 0; f (0) = 1 1 + 2−1/x 2 1 77. f (x) = where meaningful; 1 + 2tan x f (x) = 1 otherwise 1 78. f (x) = where meaningful; 1 + 21/ sin x f (x) = 0 otherwise 2 101 102 CHAPTER 2 Prelude to Calculus CHAPTER 2: REVIEW Understanding: Concepts and Definitions Refer to the listed pages to review the concepts and definitions in this chapter that you need to understand. Section Pages 2.1 The relation between secant lines and tangent lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 The difference quotient of a function f at a point x = a . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 The slope of a tangent line as a limit of difference quotients . . . . . . . . . . . . . . . . . . . . . . . . . 58 The slope formula for the tangent line at a point on a parabola . . . . . . . . . . . . . . . . . . . . . . 58–59 The relation between tangent and normal lines to a curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.2 The slope at (a, f (a)) as a limit as either h → 0 or x → a . . . . . . . . . . . . . . . . . . . . . . . . . 64 The idea of the limit of f (x) as x → a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 The constant, sum, product, quotient, and root laws of limits . . . . . . . . . . . . . . . . . . . . . . . . 68–69 The substitution law and limits of compositions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71 The four-step process for finding slope-predictor functions . . . . . . . . . . . . . . . . . . . . . . . . . . 72 2.3 The basic trigonometric limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 The squeeze law of limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Right-hand and left-hand limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79–80 The relation between one-sided and two-sided limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Existence of tangent lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Infinite limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 The precise definition of the limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 2.4 Continuity of a function at a point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Removable discontinuities of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Continuity of combinations, polynomials and rational functions . . . . . . . . . . . . . . . . . . . . . 93 Continuity of trigonometric functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Continuity of compositions of continuous functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 Continuity of a function on a closed interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 The intermediate value property of continuous functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Existence of solutions of equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Objectives: Methods and Techniques Work the listed problems in each section to practice the methods and techniques in this chapter that you need to master. Section Problems 2.1 Finding the equation of the tangent line at a point on a parabola . . . . . . . . . . . . . . . . . . . . . 9, 11 Find the point(s) on a curve where the tangent line is horizontal . . . . . . . . . . . . . . . . . . . . . 17, 21 Finding equations of both tangent and normal lines to a curve . . . . . . . . . . . . . . . . . . . . . . . 25, 27 Solving applied problems by finding high points on parabolas . . . . . . . . . . . . . . . . . . . . . . . 29, 31 Numerically investigating the slope of a tangent line at a point . . . . . . . . . . . . . . . . . . . . . . 37, 41, 45 2.2 Using limit laws to evaluate limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 7, 11 Finding limits of quotients after algebraic simplification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21, 25, 31, 35 Using the four-step process to find a slope-predictor function . . . . . . . . . . . . . . . . . . . . . . . 37, 41, 45 Investigating a limit numerically . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47, 49, 55 2.3 Using limit laws to evaluate trigonometric limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 3, 9, 11, 13, 25 Using the one-sided limit laws to evaluate limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29, 35, 39, 43, 45 Determining behavior where one-sided limits fail to exist . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 51, 55 Using the precise definition to establish a limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75, 79 2.4 Using limit laws to establish continuity of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 7, 9, 11, 13 Determining where a given function is continuous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17, 21, 23, 25, 31 Determining whether or not a discontinuity is removable . . . . . . . . . . . . . . . . . . . . . . . . . . . 37, 39, 43, 45, 47 Applying the intermediate value property to locate solutions . . . . . . . . . . . . . . . . . . . . . . . . 53, 55 Numerical investigation of continuity at a given point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73, 75 102 Chapter 2 Miscellaneous Problems 103 MISCELLANEOUS PROBLEMS Apply the limit laws to evaluate the limits in Problems 1 through 40 or to show that the indicated limit does not exist, as appropriate. 1. lim (x 2 − 3x + 4) 2. lim (3 − x + x 3 ) 3. lim (4 − x ) 4. lim (x + x − 1) x→0 x→−1 2 10 2 x→2 x→1 1 + x2 x→2 1 − x 2 6. lim x2 − 1 1−x 8. lim 5. lim 7. lim x→1 x→3 t→−3 t + 6t + 9 9 − t2 4x − x 3x + x 2 2x 2 + 1 12. lim x→2 2x 5x + 1 3/4 13. lim x→3 x2 − 8 √ x +2−3 15. lim x→7 x −7 x→−4 19. lim √ + x→2 13 + x x +4 23. lim x→2+ 16. lim x − x→1+ 1 3 4 − x2 18. lim x→1+ 2−x x −4 21. lim x→4+ |x − 4| x2 20. lim x→−2− 22. lim x→3− 27. lim x x −3 28. lim 29. lim x +1 (x − 1)3 30. lim x→1− x→1− x→5+ f (x) = 3x − x 2 + |2x + 3| at the points where a tangent line exists. Find the point (or points) where no tangent line exists. Sketch the graph of f . 55. Write equations of the two lines through (3, 4) that are tangent to the parabola y = x 2 . (Suggestion: Let (a, a 2 ) denote either point of tangency; first solve for a.) 56. Write an equation for the circle with center (2, 3) that is tangent to the line with equation x + y + 3 = 0. In Problems 57 through 60, explain why each function is continuous wherever it is defined by the given formula. For each point a where f is not defined by the formula, tell whether a value can be assigned to f (a) in such a way as to make f continuous at a. 1−x 1 − x2 x −2 x 2 − 3x + 2 59. f (x) = x2 + x − 2 x 2 + 2x − 3 25 − x 2 x 2 − 10x + 25 61. Apply the intermediate value property of continuous functions to prove that the equation x 5 + x = 1 has a solution. 1−x (2 − x)2 2 x − 1 60. f (x) = 2 x −1 58. f (x) = tan 2x tan 3x 63. Show that there is a number x between 0 and π/2 such that x = cos x. 36. lim 1 − cos 3x 2x 64. Show that there is a number x between π/2 and π such that tan x = −x. (Suggestion: First sketch the graphs of y = tan x and y = −x.) x→0 x √ sin x 37. lim 1 − cos 3x 2x 2 38. lim x 3 cot x csc x 39. lim sec 2x tan 2x x 40. lim x 2 cot2 3x x→0 x +1 x −1 57. f (x) = 35. lim x→0 x x +1 34. lim sin 3x sin 2x x→0+ 52. f (x) = 62. Apply the intermediate value property of continuous functions to prove that the equation x 3 − 4x 2 + 1 = 0 has three different solutions. 32. lim x→0 1 x tan 5x x→0 x sin 3x x→0 x 33. lim 1 2x + 1 54. Find a slope-predictor formula for the graph x 24. lim x→−3 (x + 3)2 x x −1 31. lim x2 − 1 50. f (x) = x2 − 9 26. lim x→1− 53. f (x) = x +2 |x + 2| x +2 (x − 2)2 x→3+ 48. f (x) = x − x 3 1 3−x 51. f (x) = x − 1−x |1 − x| 25. lim x→2 49. f (x) = x4 − 1 14. lim 2 x→1 x + 2x − 3 − 4 − 4x + 44. f (x) = 1 − 2x − 3x 2 x x 2 46. f (x) = − 3 4 47. f (x) = 2x 2 + 3x x→0 17. lim 43. f (x) = 3x 2 + 4x − 5 In Problems 47 through 53, use the “four-step process” of Section 2.3 to find a slope-predictor formula for the graph y = f (x). x +2 x2 + x − 2 10. lim x→3 1 42. f (x) = x − 5x 2 3 11. lim (x 2 − 1)2/3 √ 41. f (x) = 3 + 2x 2 45. f (x) = (x − 1)(2x − 1) 2x x2 − x − 3 x→−2 2 9. lim 17 In Problems 41 through 46, apply your knowledge of lines tangent to parabolas (Section 2.1) to write a slope-predictor formula for the given curve y = f (x). Then write an equation for the line tangent to y = f (x) at the point (1, f (1)). x→0 x→0 x→0 65. Find how many straight lines through the point (12, 15 ) are 2 normal to the graph of y = x 2 and find the slope of each. (Suggestion: The cubic equation you should obtain has one root evident by inspection.) 103 104 CHAPTER 2 Prelude to Calculus 66. A circle of radius r is dropped into the parabola y = x 2 . If r is too large, the circle will not fall all the way to the bottom; if r is sufficiently small, the circle will touch the parabola at its vertex (0, 0). (See Fig. 2.MP.1.) Find the largest value of r so that the circle will touch the vertex of the parabola. y = x2 FIGURE 2.MP.1 If the circle is too large, it cannot touch the bottom of the parabola (Problem 66). PHOTO CREDITS p. 53 (top left) Getty Images, Inc.-Hulton Archive Photos; (bottom left) Courtesy of International Business Machines Corporation. Unauthorized use not permitted. (right) Navy Visual News Service/U.S. Navy News Photo 104 3 The Derivative I saac Newton was born in a rural English farming village on Christmas Day in 1642, three months after his father’s death. When the boy was three, his mother remarried and left him with his grandmother. Nothing known about his childhood and early schooling hinted that his life and work would Isaac Newton (1642–1727) constitute a turning point in the history of humanity. But due to the influence of an uncle who suspected hidden potential in young Isaac, Newton was able to enter Cambridge University in 1661. During the years 1665 and 1666, when Cambridge closed because of the bubonic plague then sweeping Europe, he returned to his country home and there laid the foundations for the three towering achievements of his scientific career—the invention of the calculus, the discovery of the spectrum of colors in light, and the theory of gravitation. Of these two years he later wrote that “in those days I was in the prime of my age of invention and minded mathematics and philosophy more than at any time since.” Indeed, his thirties were devoted more to smoky chemical (and even alchemical) experiments than to serious mathematical investigations. In his forties, while a mathematics professor at Cambridge, Newton wrote the Principia Mathematica (1687), perhaps the single most influential scientific treatise ever published. In it he applied the concepts of the calculus to explore the workings of the universe, including the motions of the earth, moon, and planets about the sun. A student is said to have remarked, “There goes the man that wrote a book that neither he nor anyone else understands.” But it established for Newton such fame that upon his death in 1727 he was buried alongside his country’s greats in Westminster Abbey with such pomp that the French philosopher Voltaire remarked, “I have seen a professor of mathematics . . . buried like a king who had done good to his subjects.” Shortly after his Cambridge graduation in 1665, Newton discovered a new method for solving an equation of the form f (x) = 0. Unlike special methods such as the quadratic formula that apply only to equations of special form, Newton’s method can be used to approximate numerical solutions of virtually any equation. In Section 3.10 we present an iterative formulation of Newton’s method that is especially adaptable to calculators and computers. There we describe how the combination of Newton’s method with modern computer graphics has led to the generation of striking fractal images associated with the science of chaos. The pictures here result from the application of a complex-number version of Newton’s method to the simple equation x 3 + 1 = 0. From Chapter 3 of Calculus, Early Transcendentals, Seventh Edition. C. Henry Edwards, David E. Penney. Copyright © 2008 by Pearson Education, Inc. All rights reserved. 105 106 CHAPTER 3 The Derivative 3.1 THE DERIVATIVE AND RATES OF CHANGE y In Section 2.1 we saw that the line tangent to the curve y = f (x) (Fig. 3.1.1) at the point P(a, f (a)) has slope y = f(x) m = m(a) = lim h→0 Slope m = f'(a) f (a + h) − f (a) h (1) provided that this limit exists. As in the slope-prediction formulas of Section 2.2, we get a new function f —the derivative of the original function f —when we replace the constant a in (1) with the independent variable x. (a, f(a)) x a FIGURE 3.1.1 The geometric motivation for the definition of the derivative. DEFINITION The Derivative The derivative of the function f is the function f defined by f (x + h) − f (x) f (x) = lim h→0 h for all x for which this limit exists. (2) y It is important to understand that when the limit in (2) is evaluated, we hold x fixed while h approaches zero. When we are specifically interested in the value f (a) of the derivative f at the number x = a, we sometimes rewrite Eq. (2) in the form y = f(x) Q(x, f (x)) f (a + h) − f (a) = f(x) − f (a) P(a, f (a)) f (a) = lim h→0 h=x −a a x=a +h FIGURE 3.1.2 The notation in Eq. (3). x f (a + h) − f (a) f (x) − f (a) = lim . x→a h x −a (3) The second limit in Eq. (3) is obtained from the first by writing x = a + h, h = x − a, and by noting that x → a as h → 0 (Fig. 3.1.2). The statement that these equivalent limits exist can be abbreviated as “ f (a) exists.” In this case we say that the function f is differentiable at x = a. The process of finding the derivative f is called differentiation of f . However it is found, the derivative f is a slope predictor for lines tangent to the graph y = f (x) of the original function f (Fig. 3.1.1). The Derivative as Slope Predictor The slope m of the line tangent to the graph y = f (x) at the point (a, f (a)) where x = a is m = f (a). (4) Application of the point-slope formula gives y − f (a) = f (a) · (x − a) (5) as an equation of this tangent line. Differentiating a given function f by direct evaluation of the limit in Eq. (3) involves carrying out four steps: 1. Write the definition in Eq. (2) of the derivative. 2. Substitute the expressions f (x + h) and f (x) as determined by the particular function f . 3. Simplify the result by algebraic methods until it is possible to . . . 4. Apply appropriate limit laws to finally evaluate the limit. In Section 2.2 we used this same “four-step process” to calculate several slopepredictor functions—that is, derivatives. The limit calculations of Examples 12 and 13 in Section 2.2—where we found the derivatives of the functions √ 1 and f (x) = x f (x) = x + x 106 The Derivative and Rates of Change SECTION 3.1 107 —illustrate algebraic simplification techniques that frequently are useful in the evaluation of derivatives directly from the definition in Eq. (2). EXAMPLE 1 First apply the definition of the derivative directly to differentiate the function x . f (x) = x +3 Then find the line tangent to the graph of f at the origin, where f (0) = 0. Solution Steps 1 and 2 above give x +h x − f (x + h) − f (x) (x + h) + 3 x + 3 f (x) = lim = lim . h→0 h→0 h h Then an algebraic simplification suggested by the common-denominator calculation a c ad − bc − ad − bc b d = bd = h h hbd yields (x + h)(x + 3) − x(x + h + 3) h(x + h + 3)(x + 3) 3h 3 = lim = lim h→0 h(x + h + 3)(x + 3) h→0 (x + h + 3)(x + 3) 3 . = lim (x + h + 3) lim (x + 3) 3 f (x) = lim 2 y= 1 h→0 x 3 y 0 −1 −2 −3 −3 y= −2 h→0 x x+3 −1 h→0 We therefore find finally that 0 x 1 2 3 FIGURE 3.1.3 The tangent line y = 13 x to the curve y = x/(x + 3) at the origin. = ( px 2 ) + (q x) + r ↓ ↓ ↓ m(x) = 2( px) + q + 0 y FIGURE 3.1.4 Termwise construction of the slope-predictor function m(x) = 2 px + q for a parabola y = px 2 + q x + r. Note that the exponent 2 in the quadratic term px 2 comes “down out front”—yielding the linear term 2 px—while the linear term q x simply yields the constant q, and the constant term r just “disappears.” f (x) = 3 3 = . (x + 3)(x + 3) (x + 3)2 Substituting a = 0, f (0) = 0, and f (0) = 13 in Eq. (5) gives the equation y = 13 x of ◗ the line tangent to the graph y = x/(x + 3) at the origin (0, 0) (Fig. 3.1.3). Even when the function f is rather simple, this four-step process for computing f directly from the definition of the derivative can be time consuming. Also, Step 3 may require considerable ingenuity. Moreover, it would be very repetitious to continue to rely on this process. To avoid tedium, we want a fast, easy, short method for computing f (x). That new method is one focus of this chapter: the development of systematic methods (“rules”) for differentiating those functions that occur most frequently. Such functions include polynomials, rational functions, the trigonometric functions sin x and cos x, and combinations of such functions. Once we establish these general differentiation rules, we can apply them formally, almost mechanically, to compute derivatives. Only rarely should we need to return to the definition of the derivative. Figure 3.1.4 illustrates the slope-predictor function for a parabola that we exhibited in Eq. (10) of Section 2.1. Restated in the language of derivatives, this is an example of a “differentiation rule.” RULE Differentiation of Quadratic Functions The derivative of the quadratic function f (x) = ax 2 + bx + c (6) f (x) = 2ax + b. (7) is the linear function 107 108 CHAPTER 3 The Derivative Note that this rule works in the same way no matter whether we denote the coefficients by a, b, and c as in Eqs. (6) and (7), or by p, q, and r as in Fig. 3.1.4. It may be instructive to derive the differentiation formula in (7) directly from the definition of the derivative: f (x + h) − f (x) h→0 h [a(x + h)2 + b(x + h) + c] − [ax 2 + bx + c] = lim h→0 h 2 2 (ax + 2ahx + ah + bx + bh + c) − (ax 2 + bx + c) = lim h→0 h 2 2ahx + ah + bh = lim h→0 h = lim (2ax + ah + b). f (x) = lim h→0 Therefore f (x) = 2ax + b. Once we know this rule, we need never again apply the definition of the derivative to differentiate a quadratic function. 60 y = 3x2 − 4x + 5 50 40 30 20 y, y' 10 y decreasing y increasing 0 y' < 0 y' > 0 −10 −20 y' = 6x − 4 −30 − 40 −5 0 5 x FIGURE 3.1.5 Note that the curve y = f (x) is falling (from left to right) where the derivative f (x) is negative, and is rising where the derivative is positive. EXAMPLE 2 (a) If f (x) = 3x 2 − 4x + 5, we can apply Eq. (7) to write the derivative immediately, without going through the four-step process: f (x) = 2 · (3x) + (−4) = 6x − 4. Figure 3.1.5 compares the graph of f with that of its derivative f . (b) Similarly, if g(t) = 2t − 5t 2 , then g (t) = (2) + 2 · (−5t) = 2 − 10t. ◗ It makes no difference what the name for the function is or whether we write x or t for the independent variable. This flexibility is valuable—in general, it is such adaptability that makes mathematics applicable to virtually every other branch of human knowledge. In any case, you should learn every differentiation rule in a form independent of the notation used to state it. We develop additional differentiation rules in Sections 3.2 through 3.4. First, however, we must introduce new notation and a new interpretation of the derivative. Differential Notation An important alternative notation for the derivative originates from the early custom of writing x in place of h (because h = x is an increment in x) and y = f (x + x) − f (x) for the resulting change (or increment) in y. The slope of the secant line K of Fig. 3.1.6 is then m sec = y f (x + x) − f (x) = , x x and the slope of the tangent line is m= y dy = lim . x→0 x dx (8) dy = f (x). dx (9) Hence, if y = f (x), we often write 108 The Derivative and Rates of Change SECTION 3.1 109 y Secant line K Q Δy P y = f(x) Δx x FIGURE 3.1.6 Origin of the dy/d x notation. (The so-called differentials dy and d x are discussed carefully in Chapter 4.) The symbols f (x) and dy/d x for the derivative of the function y = f (x) are used almost interchangeably in mathematics and its applications, so you need to be familiar with both versions of the notation. You also need to know that dy/d x is a single symbol representing the derivative; it is not the quotient of two separate quantities dy and d x. EXAMPLE 2 (Continued) If y = ax 2 + bx + c, then the derivative in Eq. (7) in differential notation takes the form dy = 2ax + b. dx Consequently, if y = 3x 2 − 4x + 5, if z = 2t − 5t 2 , then then dy = 6x − 4; dx dz = 2 − 10t. dt ◗ The letter d in the notation dy/d x stands for the word “differential.” Whether we write dy/d x or dz/dt, the dependent variable appears “upstairs” and the independent variable “downstairs.” Rates of Change The derivative of a function serves as a slope predictor for straight lines tangent to the graph of that function. Now we introduce the equally important interpretation of the derivative of a function as the rate of change of that function with respect to the independent variable. We begin with the instantaneous rate of change of a function whose independent variable is time t. Suppose that Q is a quantity that varies with time t, and write Q = f (t) for the value of Q at time t. For example, Q might be • • • • • • The size of a population (such as kangaroos, people, or bacteria); The number of dollars in a bank account; The volume of a balloon being inflated; The amount of water in a reservoir with variable inflow and outflow; The amount of a chemical product produced in a reaction; or The distance traveled t hours after the beginning of a journey. The change in Q from time t to time t + t is the increment Q = f (t + t) − f (t). 109 110 CHAPTER 3 The Derivative The average rate of change of Q (per unit of time) is, by definition, the ratio of the change Q in Q to the change t in t. Thus it is the quotient Q f (t + t) − f (t) = t t (10) illustrated in Fig. 3.1.7. Q Slope: ΔQ Δt (t + Δ t, f(t + Δ t)) Δ Q = f(t + Δ t) − f (t) (the change in Q) (t, f (t)) Q Δt (the change in t) t t + Δt t Q = f(t) dQ , dt the instantaneous rate of change of Q at t Slope: (t, f(t)) Q = f(t) t FIGURE 3.1.8 The relation between the tangent line at (t, f (t)) and the instantaneous rate of change of f at t. FIGURE 3.1.7 Average rate of change as a slope. We define the instantaneous rate of change of Q (per unit of time) to be the limit of this average rate as t → 0. That is, the instantaneous rate of change of Q is Q f (t + t) − f (t) = lim . t→0 t t→0 t lim But the right-hand limit in Eq. (11) is simply the derivative f (t). So we see that the instantaneous rate of change of Q = f (t) is the derivative dQ = f (t). dt Q Slope dQ > 0: curve rising dt Q increasing t t FIGURE 3.1.9 Quantity increasing—derivative positive. (12) To interpret intuitively the concept of instantaneous rate of change, think of the point P(t, f (t)) moving along the graph of the function Q = f (t). As Q changes with time t, the point P moves along the curve. But suppose that suddenly, at the instant t, the point P begins to follow a straight-line path—like a whirling particle suddenly cut loose from its string. Then the new path of P would appear as in Fig. 3.1.8. The dashed curve in the figure corresponds to the “originally planned” behavior of Q (before P decided to fly off along the straight-line path). But the straight-line path of P (of constant slope) corresponds to the quantity Q “changing at a constant rate.” Because the straight line is tangent to the graph Q = f (t), we can interpret d Q/dt as the instantaneous rate of change of the quantity Q at the instant t: The instantaneous rate of change of Q = f (t) at time t is equal to the slope of the line tangent to the curve Q = f (t) at the point (t, f (t)). We can draw additional important conclusions. Because a positive slope corresponds to a rising tangent line and a negative slope corresponds to a falling tangent line (as in Figs. 3.1.9 and 3.1.10), we say that Q Slope dQ < 0: curve falling dt Q Q decreasing t t FIGURE 3.1.10 Quantity decreasing—derivative negative. 110 (11) Q dQ > 0; dt dQ < 0. is decreasing at time t if dt is increasing at time t if (13) NOTE The meaning of the phrase “Q = f (t) is increasing over (or during) the time interval from t = a to t = b” should be intuitively clear. The expressions in (13) give us a way to make precise what we mean by “Q = f (t) is increasing at time t”—that is, at the instant t. Note also that the fact that a function is increasing at some instant does The Derivative and Rates of Change SECTION 3.1 111 not necessarily imply that it continues to increase throughout some interval of time; this question is discussed in Section 4.3. EXAMPLE 3 The cylindrical tank in Fig. 3.1.11 has a vertical axis and is initially filled with 600 gal of water. This tank takes 60 min to empty after a drain in its bottom is opened. Suppose that the drain is opened at time t = 0. Suppose also that the volume V of water remaining in the tank after t minutes is V (t) = 16 (60 − t)2 = 600 − 20t + 16 t 2 Volume V(t) Rate V'(t) FIGURE 3.1.11 The draining tank of Example 3. gallons. Find the instantaneous rate at which water is flowing out of the tank at time t = 15 (min) and at time t = 45 (min). Also find the average rate at which water flows out of the tank during the half hour from t = 15 to t = 45. Solution The instantaneous rate of change of the volume V (t) of water in the tank is given by the derivative dV = −20 + 13 t. dt At the instants t = 15 and t = 45 we obtain and V (15) = −20 + 1 3 · 15 = −15 V (45) = −20 + 1 3 · 45 = −5. The units here are gallons per minute (gal/min). The fact that V (15) and V (45) are negative is consistent with the observation that V is a decreasing function of t (as t increases, V decreases). One way to indicate this is to say that after 15 min, the water is flowing out of the tank at 15 gal/min; after 45 min, the water is flowing out at 5 gal/min. The instantaneous rate of change of V at t = 15 is −15 gal/min, and the instantaneous rate of change of V at t = 45 is −5 gal/min. We could have predicted the units, because V /t is a ratio of gallons to minutes, and therefore its limit V (t) = d V /dt must be expressed in the same units. During the time interval of length t = 30 min from time t = 15 to time t = 45, the average rate of change of the volume V (t) is V (45) − V (15) V = t 45 − 15 1 (60 − 45)2 − 16 (60 − 15)2 −300 6 = = . 45 − 15 30 Each numerator in the last equation is measured in gallons—this is especially apparent when you examine the second numerator—and each denominator is measured in minutes. Hence the ratio in the last fraction is a ratio of gallons to minutes, so the average rate of change of the volume V of water in the tank is −10 gal/min. Thus the average rate of flow of water out of the tank during this half-hour interval is 10 gal/min. ◗ Our examples of functions up to this point have been restricted to those with formulas or verbal descriptions. Scientists and engineers often work with tables of values obtained from observations or experiments. Example 4 shows how the instantaneous rate of change of such a tabulated function can be estimated. EXAMPLE 4 The table in Fig. 3.1.12 gives the U.S. population P (in millions) in the nineteenth century at 10-year intervals. Estimate the instantaneous rate of population growth in 1850. Solution We take t = 0 (yr) in 1800, so t = 50 corresponds to the year 1850. In Fig. 3.1.13 we have plotted the given data and then added a freehand sketch of a smooth curve that fits these data. We can hope that this curve fitting the data is a good approximation to the true graph of the unknown function P = f (t). The instantaneous rate of change d P/dt in 111 112 CHAPTER 3 The Derivative t Year U.S. Population (Millions) 0 10 20 30 40 50 60 70 80 90 100 1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 5.3 7.2 9.6 12.9 17.1 23.2 31.4 38.6 50.2 63.0 76.2 P (millions) 80 60 40 (50, 23.2) 20 1800 FIGURE 3.1.12 Data for Example 4. 36 50 1850 1900 t (year) FIGURE 3.1.13 A smooth curve that fits the data of Fig. 3.1.12 well (Example 4). 1850 is then the slope of the tangent line at the point (50, 23.2). We draw the tangent line as accurately as we can by visual inspection and then measure the base and height of the triangle in Fig. 3.1.13. In this way we approximate the slope of the tangent at t = 50 as dP 36 ≈ = 0.72 dt 50 millions of people per year (in 1850). Although there was no national census in 1851, we would expect the U.S. population then to have been approximately 23.2 + 0.7 = ◗ 23.9 million. Velocity and Acceleration x=0 x = f (t) FIGURE 3.1.14 The particle in motion is at the point x = f (t) at time t. Suppose that a particle moves along a horizontal straight line, with its location x at time t given by its position function x = f (t). Thus we make the line of motion a coordinate axis with an origin and a positive direction; f (t) is merely the x-coordinate of the moving particle at time t (Fig. 3.1.14). Think of the time interval from t to t + t. The particle moves from position f (t) to position f (t + t) during this interval. Its displacement is then the increment x = f (t + t) − f (t). We calculate the average velocity of the particle during this time interval exactly as we would calculate average speed on a long motor trip: We divide the distance by the time to obtain an average speed in miles per hour. In this case we divide the displacement of the particle by the elapsed time to obtain the average velocity v= f (t + t) − f (t) x = . t t (14) (The overbar is a standard symbol that usually connotes an average of some sort.) We define the instantaneous velocity v of the particle at the time t to be the limit of the average velocity v as t → 0. That is, x f (t + t) − f (t) = lim . t→0 t t→0 t v = lim 112 (15) The Derivative and Rates of Change SECTION 3.1 113 We recognize the limit on the right in Eq. (15)—it is the definition of the derivative of f at time t. Therefore, the velocity of the moving particle at time t is simply v= dx = f (t). dt (16) Thus velocity is instantaneous rate of change of position. The velocity of a moving particle may be positive or negative, depending on whether the particle is moving in the positive or negative direction along the line of motion. We define the speed of the particle to be the absolute value |v| of the velocity. EXAMPLE 5 Figure 3.1.15 shows a car moving along the (horizontal) x-axis. Suppose that its position (in feet) at time t (in seconds) is given by x(t) = 5t 2 + 100. =0 0 = 100 x = 100 x = 600 x (ft) FIGURE 3.1.15 The car of Example 5. Then its velocity at time t is v(t) = x (t) = 10t. y = y (t) y=0 Time t Because x(0) = 100 and v(0) = 0, the car starts at time t = 0 from rest—v(0) = 0— at the point x = 100. Substituting t = 10, we see that x(10) = 600 and v(10) = 100, so after 10 s the car has traveled 500 ft (from its starting point x = 100), and its speed ◗ then is 100 ft/s. Ground level Vertical Motion FIGURE 3.1.16 Vertical motion with position function y(t). In the case of vertical motion—such as that of a ball thrown straight upward—it is common to denote the position function by y(t) rather than by x(t). Typically, y(t) denotes the height above the ground at time t, as in Fig. 3.1.16. But velocity is still the derivative of position: dy . dt Upward motion with y increasing corresponds to positive velocity, v > 0 (Fig. 3.1.17). Downward motion with y decreasing corresponds to negative velocity, v < 0. The case of vertical motion under the influence of constant gravity is of special interest. If a particle is projected straight upward from an initial height y0 (ft) above the ground at time t = 0 (s) and with initial velocity v0 (ft/s) and if air resistance is negligible, then its height y (in feet above the ground) at time t is given by a formula known from physics, v(t) = y y increasing dy = >0 dt y decreasing dy = <0 dt FIGURE 3.1.17 Upward motion and downward motion. y(t) = − 12 gt 2 + v0 t + y0 . (17) Here g denotes the acceleration due to the force of gravity. Near the surface of the earth, g is nearly constant, so we assume that it is exactly constant, and at the surface of the earth, g ≈ 32 ft/s2 , or g ≈ 9.8 m/s2 . If we differentiate y with respect to time t, we obtain the velocity of the particle at time t: v(t) = dy = −gt + v0 . dt (18) 113 114 CHAPTER 3 The Derivative The acceleration of the particle is defined to be the instantaneous time rate of change (derivative) of its velocity: a= y (19) Your intuition should tell you that a body projected upward in this way will reach its maximum height at the instant that its velocity becomes zero—when v(t) = 0. (We shall see in Section 3.5 why this is true.) t=? y=? =0 Start: t=0 y0 = 0 0 = 96 dv = −g. dt EXAMPLE 6 Find the maximum height attained by a ball thrown straight upward from the ground with initial velocity v0 = +96 ft/s. Also find the velocity with which it hits the ground upon its return. Impact: t=? y=0 =? FIGURE 3.1.18 Data for the ball of Example 6. Solution To begin the solution of a motion problem such as this, we sketch a diagram like Fig. 3.1.18, indicating both the given data and the data that are unknown at the time instants in question. Here we focus on the time t = 0 when the ball leaves the ground (y = 0), the unknown time when it reaches its maximum height with velocity v = 0, and the unknown time when it returns to the ground. We begin by substituting y0 = 0, v0 = 96, and g = 32 in Eq. (17). Then the height of the ball at time t (so long as it remains aloft) is given by y(t) = −16t 2 + 96t. Then differentiation gives its velocity at time t, 200 v(t) = y (t) = −32t + 96 v = −32t + 96 y= 150 −16t2 + 96t (see Fig. 3.1.19). The ball attains its maximum height when v = 0; that is, when v(t) = −32t + 96 = 0. 100 y, v 50 y inc. v>0 0 −50 −2 0 2 This occurs when t = 3 (s). Substituting t = 3 in the height function y(t) gives the maximum height of the ball, y dec. v<0 t 4 ymax = y(3) = −16 · (3)2 + 96 · (3) = 144 6 8 FIGURE 3.1.19 Note that the ball is rising when its velocity v > 0, falling when v < 0, and is at its apex when v = 0. (ft). The ball returns to the ground when y(t) = 0. The equation y(t) = −16t 2 + 96t = −16t (t − 6) = 0 has the two solutions t = 0 and t = 6. Thus the ball returns to the ground at time t = 6. The velocity with which it strikes the ground is v(t) = (−32)(6) + 96 = −96 (ft/s). ◗ Other Rates of Change The derivative of any function—not merely a function of time—may be interpreted as its instantaneous rate of change with respect to the independent variable. If y = f (x), then the average rate of change of y (per unit change in x) on the interval [x, x + x] is the quotient y f (x + x) − f (x) = . x x The instantaneous rate of change of y with respect to x is the limit, as x → 0, of the average rate of change. Thus the instantaneous rate of change of y with respect to x is y dy = = f (x). x→0 x dx lim 114 (20) The Derivative and Rates of Change SECTION 3.1 115 Example 7 illustrates the fact that a dependent variable may sometimes be expressed as two different functions of two different independent variables. The derivatives of these functions are then rates of change of the dependent variable with respect to the two different independent variables. Δx x Δx EXAMPLE 7 The area of a square with edge length x centimeters is A = x 2 , so the derivative of A with respect to x, dA = 2x, dx A = x2 (21) is the rate of change of its area A with respect to x. (See the computations in Fig. 3.1.20.) The units of d A/d x are square centimeters per centimeter. Now suppose that the edge length of the square is increasing with time: x = 5t, with time t in seconds. Then the area of the square at time t is x A = (5t)2 = 25t 2 . FIGURE 3.1.20 The square of Example 7: The derivative of A with respect to t is A + A = (x + x)2 ; dA = 2 · 25t = 50t; dt A = 2xx + (x)2 ; A = 2x + x; x dA = 2x. dx (22) this is the rate of change of A with respect to time t, with units of square centimeters per second. For instance, when t = 10 (so x = 50), the values of the two derivatives of A in Eqs. (21) and (22) are d A = 2 · 50 = 100 d x x=50 (cm2 /cm) d A = 50 · 10 = 500 dt t=10 (cm2 /s). and Thus A is increasing at the rate of 100 cm2 per cm increase in x, and at the rate of 500 cm2 per second increase in t. ◗ 800 700 600 500 400 y 300 200 100 0 −100 −200 −4 The notation d A/dt for the derivative suffers from the minor inconvenience of not providing a “place” to substitute a particular value of t, such as t = 10. The last lines of Example 7 illustrate one way around this difficulty. Just as we can speak of whether the quantity Q(t) is increasing or decreasing at time t = a—according as Q (a) > 0 or Q (a) < 0—we can ask whether the function y = f (x) is an increasing or decreasing function of x. Thinking of rising tangent lines with positive slopes, and falling tangent lines with negative slopes, we say in analogy with (13) that y = f(x) −2 0 2 y = f '(x) 4 y is increasing at the point x = a y is decreasing at the point x = a 6 if f (a) > 0; if f (a) < 0. x FIGURE 3.1.21 Correspondence between the function graph y = f (x) and the derivative graph y = f (x). EXAMPLE 8 Figure 3.1.21 shows the graphs y = f (x) of a function and y = f (x) of its derivative. Observe that • • • y = f (x) has a horizontal tangent line at points where f (x) = 0; f (x) is increasing on open intervals where f (x) > 0; and f (x) is decreasing on open intervals where f (x) < 0. ◗ 115 116 CHAPTER 3 The Derivative 3.1 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The derivative of the function f is the function f with the rule f (x + h) − f (x) h→0 h for those values of x for which the limit exists. If f (a) exists, then there is a straight line tangent to the graph of f at the point P(a, f (a)), and its slope is f (a). If p, q, and r are constants and f (x) = px 2 + q x + r , then f (x) = 2 px + q. dy as an alternative notation for f (x); If y = f (x), then it is acceptable to write dx dy = f (x). that is: If y = f (x), then dx If Q = Q(t) is a function of time t, then the average rate of change of Q over Q(t + t) − Q(t) the time interval [t, t + t] is . t If Q = Q(x) is a function of x, then the instantaneous rate of change of Q with Q(x + h) − Q(x) respect to x is Q (x) = lim . h→0 h If a particle moves along a straight line with position x(t) at time t and velocity v(t) at time t, then v (t) = x(t). If a particle moves along a straight line with velocity v(t) at time t, then its acceleration a(t) at time t is defined to be a(t) = v (t). If Q = f (t) is a function of time t, then Q is increasing at the instant t if f (t) > 0. If y = f (x) is a function of x, then y is decreasing at x provided that f (x) < 0. f (x) = lim 2. 3. 4. 5. 6. 7. 8. 9. 10. 3.1 CONCEPTS: QUESTIONS AND DISCUSSION 1. The slope line in Fig. 3.1.5 looks as if it might be tangent to the parabola. Is it? If not, what’s a simple way you could alter the equation of the parabola—without changing its slope line—in order to ensure that the line will be tangent to the altered parabola? 2. When a ball is tossed straight upward, it may appear to hover at the apex of its trajectory for a brief period of time. Does it? 3. You are pulled over by a policeman who claims that you did not stop properly at a stop sign. You argue that as you braked your car, its velocity was zero at a certain instant before you removed your foot from the brake pedal and proceeded through the intersection. The policeman replies that you nevertheless did not come to a full stop—that he is certain your velocity did not remain zero for even a hundredth of a second. What’s the cause of this disagreement? Explain it with such convincing clarity that the judge will let you off without you paying a fine. 4. The ball of Example 6 took the same amount of time to rise from the ground to its highest point as to fall back to the ground. Is this always the case for a ball governed by Eqs. (17) and (18) of this section? Suggestion: In lieu of a blizzard of algebra, think about the symmetry of the parabola in Fig. 3.1.19. 3.1 PROBLEMS In Problems 1 through 10, find the indicated derivative by using the differentiation rule in Eqs. (6) and (7): If 116 f (x) = ax 2 + bx + c, then f (x) = 2ax + b. 1. f (x) = 4x − 5; find f (x). 2. g(t) = 100 − 16t 2 ; find g (t). 3. h(z) = z(25 − z); find h (z). The Derivative and Rates of Change SECTION 3.1 4. f (x) = 16 − 49x; find f (x). 5. y = 2x 2 + 3x − 17; find dy/d x. 6. x = 16t − 100t 2 ; find d x/dt. 7. z = 5u 2 − 3u; find dz/du. 8. v = 5y(100 − y); find dv/dy. 9. x = −5y 2 + 17y + 300; find d x/dy. 10. u = 7t 2 + 13t; find du/dt. In Problems 11 through 20, apply the definition of the derivative (as in Example 1) to find f (x). 11. f (x) = 2x − 1 12. f (x) = 2 − 3x 13. f (x) = x 2 + 5 14. f (x) = 3 − 2x 2 15. f (x) = 1 2x + 1 16. f (x) = 17. f (x) = √ 2x + 1 18. f (x) = √ 1 3−x 1 x +1 In Problems 21 through 25, the position function x = f (t) of a particle moving in a horizontal straight line is given. Find its location x when its velocity v is zero. 21. x = 100 − 16t 2 22. x = −16t 2 + 160t + 25 23. x = −16t 2 + 80t − 1 24. x = 100t 2 + 50 33. Figure 3.1.25 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 0 x 5 0 x FIGURE 3.1.24 FIGURE 3.1.25 34. Figure 3.1.26 35. Figure 3.1.27 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 0 x 5 FIGURE 3.1.26 x +1 20. f (x) = x −1 x 19. f (x) = 1 − 2x 32. Figure 3.1.24 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 25. x = 100 − 20t − 5t 2 5 0 x 5 FIGURE 3.1.27 0 x 5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 0 x FIGURE 3.1.28(a) FIGURE 3.1.28(b) In Problems 30 through 35 (Figs. 3.1.22 through 3.1.27), match the given graph of the function f with that of its derivative, which appears among those in Fig. 3.1.28, parts (a) through ( f ). 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 30. Figure 3.1.22 31. Figure 3.1.23 FIGURE 3.1.28(c) 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 In Problems 26 through 29, the height y(t) (in feet at time t seconds) of a ball thrown vertically upward is given. Find the maximum height that the ball attains. 26. y = −16t 2 + 160t 27. y = −16t 2 + 64t 28. y = −16t 2 + 128t + 25 29. y = −16t 2 + 96t + 50 FIGURE 3.1.22 0 x 5 FIGURE 3.1.23 0 x 5 117 0 x 0 x FIGURE 3.1.28(e) 5 0 x 5 5 FIGURE 3.1.28(d) 5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 0 x 5 FIGURE 3.1.28(f) 117 118 CHAPTER 3 The Derivative estimate its speed (in miles per hour) when t = 20 and again when t = 40. 36. The Celsius temperature C is given in terms of the Fahrenheit temperature F by C = 59 (F − 32). Find the rate of change of C with respect to F and the rate of change of F with respect to C. 37. Find the rate of change of the area A of a circle with respect to its circumference C. 38. A stone dropped into a pond at time t = 0 s causes a circular ripple that travels out from the point of impact at 5 m/s. At what rate (in square meters per second) is the area within the circle increasing when t = 10? 39. A car is traveling at 100 ft/s when the driver suddenly applies the brakes (x = 0, t = 0). The position function of the skidding car is x(t) = 100t − 5t 2 . How far and for how long does the car skid before it comes to a stop? 40. A water bucket containing 10 gal of water develops a leak at time t = 0, and the volume V of water in the bucket t seconds later is given by t 2 V (t) = 10 1 − 100 until the bucket is empty at time t = 100. (a) At what rate is water leaking from the bucket after exactly 1 min has passed? (b) When is the instantaneous rate of change of V equal to the average rate of change of V from t = 0 to t = 100? 41. A population of chipmunks moves into a new region at time t = 0. At time t (in months), the population numbers P(t) = 100[1 + (0.3)t + (0.04)t 2 ]. (a) How long does it take for this population to double its initial size P(0)? (b) What is the rate of growth of the population when P = 200? 42. The following data describe the growth of the population P (in thousands) of Gotham City during a 10-year period. Use the graphical method of Example 4 to estimate its rate of growth in 1989. Year 1984 1986 1988 1990 1992 1994 P 265 293 324 358 395 427 43. The following data give the distance x in feet traveled by an accelerating car (that starts from rest at time t = 0) in the first t seconds. Use the graphical method of Example 4 to t 0 10 20 30 40 50 60 x 0 224 810 1655 2686 3850 5109 In Problems 44 through 49, use the fact (proved in Section 3.2) that the derivative of y = ax 3 + bx 2 + cx + d is dy/d x = 3ax 2 + 2bx + c. 44. Prove that the rate of change of the volume V of a cube with respect to its edge length x is equal to half the surface area A of the cube (Fig. 3.1.29). 45. Show that the rate of change of the volume V of a sphere with respect to its radius R is equal to its surface area S (Fig. 3.1.30). 46. The height h of a certain cylinder whose height changes is always twice its radius r . Show that the rate of change of its volume V with respect to r is equal to its total surface area S (Fig. 3.1.31). 47. A spherical balloon with an initial radius r of 5 in. begins to leak at time t = 0, and its radius t seconds later is r = (60 − t)/12 in. At what rate (in cubic inches per second) is air leaking from the balloon when t = 30? 48. The volume V (in liters) of 3 g of CO2 at 27◦ C is given in terms of its pressure p (in atmospheres) by the formula V = 1.68/ p. What is the rate of change of V with respect to p when p = 2 (atm)? (Suggestion: Use the fact that the derivative of f (x) = c/x is f (x) = −c/x 2 if c is a constant; you can establish this by using the definition of the derivative.) 49. As a snowball with an initial radius of 12 cm melts, its radius decreases at a constant rate. It begins to melt when t = 0 (h) and takes 12 h to disappear. (a) What is its rate of change of volume when t = 6? (b) What is its average rate of change of volume from t = 3 to t = 9? 50. A ball thrown vertically upward at time t = 0 (s) with initial velocity 96 ft/s and with initial height 112 ft has height function y(t) = −16t 2 + 96t + 112. (a) What is the maximum height attained by the ball? (b) When and with what impact speed does the ball hit the ground? 51. A spaceship approaching touchdown on the planet Gzyx has height y (meters) at time t (seconds) given by y = 100 − 100t + 25t 2 . When and with what speed does it hit the ground? h r x r x x FIGURE 3.1.29 The cube of Problem 44—volume V = x 3 , surface area S = 6x 2 . 118 FIGURE 3.1.30 The sphere of Problem 45—volume V = 43 πr 3 , surface area S = 4πr 2 . FIGURE 3.1.31 The cylinder of Problem 46—volume V = πr 2 h, surface area S = 2πr 2 + 2πr h. Basic Differentiation Rules SECTION 3.2 52. The population (in thousands) of the city Metropolis is given by P(t) = 100[1 + (0.04)t + (0.003)t 2 ], with t in years and with t = 0 corresponding to 1980. (a) What was the rate of change of P in 1986? (b) What was the average rate of change of P from 1983 to 1988? 53. Suppose that during the 1990s the population P of a small city was given by P(t) = 10 + t − 0.1t 2 + 0.006t 3 (with t in years and P in thousands). Taking t = 0 on January 1, 1990, find the time(s) during the 1990s at which the instantaneous rate of change of this population was equal to its average rate of change during the whole decade. (Use the differentiation formulas given in the instructions for Problems 44–49.) Problems 54 through 60 involve the left-hand and right-hand derivatives of f at a that are defined by 119 54. (a) Find f − (0) and f + (0) given f (x) = |x|. (b) The function f (x) = |12x − 101| is differentiable except at a single point. What is this point, and what are the values of its lefthand and right-hand derivatives of f there? 55. Sketch the graph of the given function f and determine if it is differentiable at x = 0: x if x < 0, (a) f (x) = 2x if x 0; 2 if x < 0, x (b) f (x) = 2x 2 if x 0. 56. Investigate the differentiability of the function f defined by 2x + 1 if x < 1, f (x) = 4x − x 2 if x 1. 57. Investigate the differentiability of the function f defined by 11 + 6x − x 2 if x < 3, f (x) = 2 x − 6x + 29 if x 3. f − (a) = lim f (a + h) − f (a) h 58. Sketch the graph of the function f (x) = x ·|x| and show that it is differentiable everywhere. Can you write a single (onepart) formula that gives the value of f (x) both for x > 0 and for x < 0? f + (a) = lim f (a + h) − f (a) , h 59. Sketch the graph of the function f (x) = x +|x|. Then investigate its differentiability. Find the derivative f (x) where it exists; find the one-sided derivatives at the points where f (x) does not exist. h→0− and h→0+ (assuming these limits exist). Then f (a) exists if and only both the left-hand and right-hand derivatives exist and f − (a) = f + (a). 60. Repeat Problem 59, except with the function f (x) = x · (x + |x|). 3.2 BASIC DIFFERENTIATION RULES Here we begin our systematic development of formal rules for finding the derivative f of the function f : f (x) = lim h→0 f (x + h) − f (x) . h (1) Some alternative notation for derivatives will be helpful. When we interpreted the derivative in Section 3.1 as a rate of change, we found it useful to employ the dependent-independent variable notation f y = f (x), x = h, y = f (x + x) − f (x). (2) This led to the “differential notation” dy y f (x + x) − f (x) = lim = lim x→0 x→0 dx x x Dx Dx f(x) = f '(x) FIGURE 3.2.1 The “differentiation machine” Dx . (3) for the derivative. When you use this notation, remember that the symbol dy/d x is simply another notation for the derivative f (x); it is not the quotient of two separate entities dy and d x. A third notation is sometimes used for the derivative f (x); it is Dx f (x). Here, think of Dx as a “machine” that operates on the function f to produce its derivative Dx f with respect to x (Fig. 3.2.1). Thus we can write the derivative 3x 2 of y = f (x) = x 3 in any of three ways: f (x) = dy = Dx x 3 = 3x 2 . dx 119 120 CHAPTER 3 The Derivative These three notations for the derivative—the function notation f (x), the differential notation dy/d x, and the operator notation Dx f (x)—are used almost interchangeably in mathematical and scientific writing, so you need to be familiar with each. The Derivative of a Constant y y=c Slope zero x FIGURE 3.2.2 The derivative of a constant-valued function is zero (Theorem 1). Our first differentiation rule says that the derivative of a constant function is identically zero. Geometry makes this obvious, because the graph of a constant function is a horizontal straight line that is its own tangent line, with slope zero at every point (Fig. 3.2.2). THEOREM 1 Derivative of a Constant If f (x) = c (a constant) for all x, then f (x) = 0 for all x. That is, dc = Dx c = 0. dx (4) Proof Because f (x + h) = f (x) = c, we see that f (x) = lim h→0 f (x + h) − f (x) c−c 0 = lim = lim = 0. h→0 h→0 h h h ◆ The Power Rule As motivation for the next rule, consider the following list of derivatives, all of which have already appeared in the text (or as problems). The first two are special cases of the formula Dx (ax 2 + bx + c) = 2ax + b. Dx x = 1 Dx x 2 = 2x = 2 · x 1 Dx x 3 = 3 · x 2 1 1 = Dx x −1 = − 2 = −1 · x −2 x x (Problem 38, Section 2.2) 1 2 = Dx x −2 = − 3 = −2 · x −3 2 x x (Problem 39, Section 2.2) Dx Dx (Problem 37, Section 2.2) √ 1 1 Dx x = Dx x 1/2 = √ = · x −1/2 2 2 x (Example 13, Section 2.2) Each of these formulas fits the simple pattern Dx x n = nx n−1 , (5) in which the exponent n is simultaneously placed before the variable and, in the exponent, is decreased by 1. Thus it appears that the blanks in the pattern Dx x = x −1 can be filled with any (single) integer you please, or even the fraction 12 . But Eq. (5)— inferred from the preceding list of derivatives—is as yet only a conjecture. Nevertheless, many discoveries in mathematics are made by detecting such patterns, then proving that they hold universally. Eventually we shall see that the formula in Eq. (5), called the power rule, is valid for all real numbers n. At this time we give a proof only for the case in which the exponent n is a positive integer. 120 Basic Differentiation Rules SECTION 3.2 121 THEOREM 2 Power Rule for a Positive Integer n If n is a positive integer and f (x) = x n , then f (x) = nx n−1 . (6) Proof For a positive integer n, the identity bn − a n = (b − a)(bn−1 + bn−2 a + bn−3 a 2 + · · · + ba n−2 + a n−1 ) is easy to verify by multiplication. Thus, if b = a, then bn − a n = bn−1 + bn−2 a + bn−3 a 2 + · · · + ba n−2 + a n−1 . b−a Because each of the n terms on the right-hand side approaches a n−1 as b → a, this tells us that lim b→a bn − a n = na n−1 b−a by various limit laws. Now let b = x + h and a = x, so that h = b − a. Then h → 0 as b → a, and hence (x + h)n − x n = nx n−1 . h→0 h f (x) = lim (7) ◆ This establishes Theorem 2. We need not always use the same symbols x and n for the independent variable and the constant exponent in the power rule. For instance, Dt t m = mt m−1 and Dz z k = kz k−1 . If it is perfectly clear what the independent variable is, the subscript may be dropped from Dx (or Dt , or Dz ), as in Example 1. EXAMPLE 1 Dx 7 = 7x 6 , Dt 17 = 17t 16 , Dz 100 = 100z 99 . ◗ The Derivative of a Linear Combination To use the power rule to differentiate polynomials, we need to know how to differentiate linear combinations. A linear combination of the functions f and g is a function of the form a f + bg where a and b are constants. It follows from the sum and product laws for limits that (8) lim [a f (x) + bg(x)] = a lim f (x) + b lim g(x) x→c x→c x→c provided that the two limits on the right in Eq. (8) both exist. The formula in Eq. (8) is called the linearity property of the limit operation. It implies an analogous linearity property of differentiation. THEOREM 3 Derivative of a Linear Combination If f and g are differentiable functions and a and b are fixed real numbers, then Dx [a f (x) + bg(x)] = a[Dx f (x)] + b[Dx g(x)]. (9) With u = f (x) and v = g(x), this takes the form d(au + bv) du dv =a +b . dx dx dx (9 ) 121 122 CHAPTER 3 The Derivative Proof The linearity property of limits immediately gives [a f (x + h) + bg(x + h)] − [a f (x) + bg(x)] Dx [a f (x) + bg(x)] = lim h→0 h f (x + h) − f (x) g(x + h) − g(x) = a lim + b lim h→0 h→0 h h = a[Dx f (x)] + b[Dx g(x)], ◆ as desired. Now take a = c and b = 0 in Eq. (9). The result is Dx [c f (x)] = cDx f (x); (10) d(cu) du =c , dx dx (10 ) alternatively, Thus the derivative of a constant multiple of a function is the same constant multiple of its derivative. EXAMPLE 2 (a) Dx (16x 6 ) = 16 · 6x 5 = 96x 5 . (b) If f (z) = 7z 3 , then f (z) = 21z 2 . d (99u 100 ) = 9900u 99 . (c) du ◗ Next, take a = b = 1 in Eq. (9). We find that Dx [ f (x) + g(x)] = [Dx f (x)] + [Dx g(x)]. (11) In differential notation, du dv d(u + v) = + . dx dx dx (11 ) Thus the derivative of the sum of two functions is the sum of their derivatives. Similarly, for differences we have du dv d(u − v) = − . dx dx dx (12) It’s easy to see that these rules generalize to sums and differences of more than two functions. For example, repeated application of Eq. (11) to the sum of a finite number of differentiable functions gives du 1 du 2 du n d(u 1 + u 2 + · · · + u n ) = + + ··· + . dx dx dx dx (13) REMARK Equation (13) tells us that, when differentiating a sum of terms, we simply differentiate each term and then add the results. EXAMPLE 3 Dx (36 + 26x + 7x 5 − 5x 9 ) = 0 + 26 · 1 + 7 · 5x 4 − 5 · 9x 8 = 26 + 35x 4 − 45x 8 . 122 ◗ Basic Differentiation Rules SECTION 3.2 123 The Derivative of a Polynomial When we apply Eqs. (10) and (13) and the power rule to the polynomial p(x) = an x n + an−1 x n−1 + · · · + a2 x 2 + a1 x + a0 (and thus differentiate termwise), we find the derivative as fast as we can write it: p (x) = nan x n−1 + (n − 1)an−1 x n−2 + · · · + 3a3 x 2 + 2a2 x + a1 . (14) With this result, it becomes a routine matter to write an equation for a line tangent to the graph of a polynomial. EXAMPLE 4 Write an equation for the straight line that is tangent to the graph of y = 2x 3 − 7x 2 + 3x + 4 at the point (1, 2). 30 20 y = 2x3 − 7x2 + 3x + 4 10 y Solution We compute the derivative as in Eq. (14): dy = 2 · 3x 2 − 7 · 2x + 3 + 0 = 6x 2 − 14x + 3. dx (1, 2) 0 −10 We substitute x = 1 in dy/d x and find that the slope of the tangent line at (1, 2) is m = −5. So the point-slope equation of the tangent line is y = −5x + 7 −20 −30 − 4 −3 −2 −1 y − 2 = −5(x − 1); 0 x 1 2 3 FIGURE 3.2.3 The graph y = 2x 3 − 7x 2 + 3x + 4 and its tangent line y = −5x + 7 at the point (1, 2). 4 that is, y = −5x + 7. A calculator- or computer-generated picture like Fig. 3.2.3 provides suggestive visual evidence of the validity of this tangent line computation. ◗ EXAMPLE 5 The volume V (in cubic centimeters) of a given sample of water varies with changing temperature T . For T between 0◦ C and 30◦ C, the relation is given almost exactly by the formula V = V0 [1 − (6.427 × 10−5 )T + (8.505 × 10−6 )T 2 − (6.790 × 10−8 )T 3 ], where V0 is the volume of the water (not ice) sample at 0◦ C. Suppose that V0 = 105 cm3 . Find both the volume and the rate of change of volume with respect to temperature when T = 20◦ C. Solution Substituting V0 = 105 = 100,000 in the given volume formula yields V (T ) = 100,000 − (6.427)T + (0.8505)T 2 − (0.00679)T 3 . Then substituting T = 20 yields V (20) ≈ 100,157.34, so the sample would expand by about 157 cm3 if heated from 0◦ C to 20◦ C. The rate of change of volume V with respect to temperature T is given by dV = −6.427 + (1.7010)T − (0.02037)T 2 , dT and substituting T = 20 here yields d V ≈ 19.45 dT T =20 (cm3 /◦ C). Thus we should expect the volume of the water sample to increase by slightly more than 19 cm3 if it is heated by 1◦ C from 20◦ C to 21◦ C. In fact, direct substitution into the original volume formula gives V (21) − V (20) ≈ 19.88. 123 124 CHAPTER 3 The Derivative Finally, we note that the average rate of change of V with respect to T on the interval 19.5 T 20.5 centered at T = 20 is V (20.5) − V (19.5) V = ≈ 19.44 T 20.5 − 19.5 (cm3 /◦ C), which is very close to the derivative d V /dT at T = 20. ◗ The Product Rule and the Quotient Rule It might be natural to conjecture that the derivative of a product f (x)g(x) is the product of the derivatives. This is false! For example, if f (x) = g(x) = x, then Dx [ f (x)g(x)] = Dx x 2 = 2x. But [Dx f (x)] · [Dx g(x)] = (Dx x) · (Dx x) = 1 · 1 = 1. In general, the derivative of a product is not merely the product of the derivatives. Theorem 4 tells us what it is. THEOREM 4 The Product Rule If f and g are differentiable at x, then f g is differentiable at x, and Dx [ f (x)g(x)] = f (x)g(x) + f (x)g (x). (15) With u = f (x) and v = g(x), this product rule takes the form d(uv) dv du =u +v . (15 ) dx dx dx When it is clear what the independent variable is, we can write the product rule even more briefly: (uv) = u v + uv . (15 ) Proof We use an “add and subtract” device. Dx [ f (x)g(x)] = lim h→0 = lim h→0 f (x + h)g(x + h) − f (x)g(x) h f (x + h)g(x + h) − f (x)g(x + h) + f (x)g(x + h) − f (x)g(x) h f (x + h)g(x + h) − f (x)g(x + h) f (x)g(x + h) − f (x)g(x) + lim h→0 h→0 h h f (x + h) − f (x) g(x + h) − g(x) lim g(x + h) + f (x) lim = lim h→0 h→0 h→0 h h = lim = f (x)g(x) + f (x)g (x). ◆ In this proof we used the sum law and product law for limits, the definitions of f (x) and g (x), and the fact that lim g(x + h) = g(x). h→0 This last equation holds because g is differentiable and therefore continuous at x (as we will see in Theorem 2 in Section 3.4). In words, the product rule says that the derivative of the product of two functions is formed by multiplying the derivative of each by the other and then adding the results. 124 Basic Differentiation Rules SECTION 3.2 125 EXAMPLE 6 Find the derivative of f (x) = (1 − 4x 3 )(3x 2 − 5x + 2) without first multiplying out the two factors. Solution Dx [(1 − 4x 3 )(3x 2 − 5x + 2)] = [Dx (1 − 4x 3 )](3x 2 − 5x + 2) + (1 − 4x 3 )[Dx (3x 2 − 5x + 2)] = (−12x 2 )(3x 2 − 5x + 2) + (1 − 4x 3 )(6x − 5) = −60x 4 + 80x 3 − 24x 2 + 6x − 5. ◗ We can apply the product rule repeatedly to find the derivative of a product of three or more differentiable functions u 1 , u 2 , . . . , u n of x. For example, D[u 1 u 2 u 3 ] = (u 1 u 2 ) u 3 + (u 1 u 2 )u 3 = (u 1 u 2 + u 1 u 2 )u 3 + u 1 u 2 u 3 = u 1 u 2 u 3 + u 1 u 2 u 3 + u 1 u 2 u 3 . Note that the derivative of each factor in the original product is multiplied by the other two factors and then the three resulting products are added. This is, indeed, the general result: D(u 1 u 2 · · · u n ) = u 1 u 2 u 3 · · · u n−1 u n + u 1 u 2 u 3 · · · u n−1 u n + · · · + u 1 u 2 u 3 · · · u n−1 u n , (16) where the sum in Eq. (16) has one term corresponding to each of the n factors in the product u 1 u 2 · · · u n . It is easy to establish this extended product rule (see Problem 62) one step at a time—next with n = 4, then with n = 5, and so forth. Our next result tells us how to find the derivative of the reciprocal of a function if we know the derivative of the function itself. THEOREM The Reciprocal Rule If f is differentiable at x and f (x) = 0, then Dx 1 f (x) . =− f (x) [ f (x)]2 (17) With u = f (x), the reciprocal rule takes the form d 1 1 du =− 2 · . dx u u dx (17 ) If there can be no doubt what the independent variable is, we can write 1 u = − 2. u u (17 ) Proof As in the proof of Theorem 4, we use the limit laws, the definition of the derivative, and the fact that a function is continuous wherever it is differentiable (by Theorem 2 of Section 3.4). Moreover, note that f (x + h) = 0 for h near zero because f (x) = 0 and f is continuous at x. (See Problem 16 in Appendix D.) Therefore 1 1 1 1 f (x) − f (x + h) = lim − = lim Dx h→0 h h→0 h f (x + h) f (x) f (x) f (x + h) f (x) 1 = − lim h→0 f (x + h) f (x) f (x + h) − f (x) lim h→0 h =− f (x) . ◆ [ f (x)]2 125 126 CHAPTER 3 The Derivative EXAMPLE 7 With f (x) = x 2 + 1 in Eq. (17), we get 2x Dx (x 2 + 1) 1 Dx =− 2 . = − 2 2 2 x +1 (x + 1) (x + 1)2 We now combine the reciprocal rule with the power rule for positive integral exponents to establish the power rule for negative integral exponents. THEOREM 5 Power Rule for a Negative Integer n If n is a negative integer, then Dx x n = nx n−1 . Proof Let m = −n, so that m is a positive integer. If x = 0 then we can apply the reciprocal rule with f (x) = x m = 0 and f (x) = mx m−1 (the latter by the power rule with positive integer exponent). This gives mx m−1 1 Dx (x m ) = − = (−m)x −m−1 = nx n−1 . = − Dx x n = Dx xm x 2m (x m )2 Thus we have established that the rule in Theorem 5 holds precisely where the function ◆ being differentiated is defined—that is, where x = 0. EXAMPLE 8 4 5x − 6x + 7 Dx = Dx 2x 2 5 2 x 2 − 3x −1 + 72 x −2 = 52 (2x) − 3(−x −2 ) + 72 (−2x −3 ) = 5x + 3 7 − 3. 2 x x ◗ The key here was to “divide out” before differentiating. Now we apply the product rule and reciprocal rule to get a rule for differentiation of the quotient of two functions. THEOREM 6 The Quotient Rule If f and g are differentiable at x and g(x) = 0, then f /g is differentiable at x and Dx f (x)g(x) − f (x)g (x) f (x) . = g(x) [g(x)]2 (18) With u = f (x) and v = g(x), this rule takes the form du dv d u v dx − u dx = (18 ) dx v v2 If it is clear what the independent variable is, we can write the quotient rule in the form u u v − uv = . (18 ) v v2 Proof We apply the product rule to the factorization f (x) 1 = f (x) · . g(x) g(x) This gives Dx 126 f (x) 1 1 = [Dx f (x)] · + f (x) · Dx g(x) g(x) g(x) g (x) f (x)g(x) − f (x)g (x) f (x) . + f (x) · − = = g(x) [g(x)]2 [g(x)]2 ◆ Basic Differentiation Rules SECTION 3.2 127 Note that the numerator in Eq. (18) is not the derivative of the product of f and g. And the minus sign means that the order of terms in the numerator is important. EXAMPLE 9 Find z (t) = dz/dt if z is given by z= 2 Solution Here, primes denote derivatives with respect to t. With t (rather than x) as the independent variable, the quotient rule gives 1 z (t) z (1 − t 3 ) (1 + t 4 ) − (1 − t 3 )(1 + t 4 ) dz = dt (1 + t 4 )2 0 −1 −2 −5 1 − t3 . 1 + t4 = z '(t) 0 t FIGURE 3.2.4 Graphs of the function z(t) of Example 9 and its derivative z (t). 5 (−3t 2 )(1 + t 4 ) − (1 − t 3 )(4t 3 ) t 6 − 4t 3 − 3t 2 = . (1 + t 4 )2 (1 + t 4 )2 Figure 3.2.4 shows computer-generated graphs of the function z(t) and its derivative z (t). Observe that z(t) is increasing on intervals where z (t) is positive and is decreasing on intervals where z (t) is negative (thus corroborating our computation of the derivative). A quick computer or calculator graph of a function and its alleged derivative will often reveal an error if one has been made. ◗ 3.2 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If y = f (x), then three acceptable notational devices for indicating the derivative dy , and Dx f (x). of f are f (x), dx 3 2. Dx (x −3/2 ) = − x −1/2 . 2 3. Dx (16x 6 ) = 22x 5 . 4. If f (x) = 2x 3 − 7x 2 + 3x + 4, then f (x) = 6x 2 − 14x + 3 + 4. dy = 2x · (x 3 − 1) + 3x 2 · (x 2 + 1). 5. If y = y(x) = (x 2 + 1) · (x 3 − 1), then dx 1 − t3 , then 6. If z = z(t) = 1 + t4 (−3t 2 ) · (1 + t 4 ) − (1 − t 3 ) · (4t 3 ) dz = . dt (1 + t 4 )2 7. If Dx (sin x) = cos x, then Dz (sin z) = cos z. 8. If Dx (sin x) = cos x, then Dx (x sin x) = x cos x + sin x. sin x 9. If Dx (sin x) = cos x, then Dx = 1. x 10. If u and v are differentiable functions of x, then the assertion that du dv d(u + v) = + dx dx dx is both notationally and mathematically correct. 3.2 CONCEPTS: QUESTIONS AND DISCUSSION 1. Theorems 2 and 5 in this section imply that the power rule Dx x n = nx n−1 holds provided that the integer n is nonzero. Does it also hold if n = 0? Can you think of a simple algebraic function whose derivative is a nonzero constant multiple of 1/x? 127 128 CHAPTER 3 The Derivative 2. Example 1 and the discussion preceding it may seem to imply that the power rule holds in the very general form D[whatever]n = n[whatever]n−1 —more precisely, Dx [ f (x)]n = n[ f (x)]n−1 . Is this true or false? When you’re confronted with a question like this, don’t just sit there. Check it out! Test the conjecture with specific choices for n and f (x)—perhaps n = 7 and f (x) = x 11 . What are the simplest choices you can use to resolve the matter? 3.2 PROBLEMS Apply the differentiation rules of this section to find the derivatives of the functions in Problems 1 through 40. 1. f (x) = 3x 2 − x + 5 3. f (x) = (2x + 3)(3x − 2) 5. h(x) = (x + 1) 7. 9. 11. 13. 2. g(t) = 1 − 3t 2 − 2t 4 4. g(x) = (2x 2 − 1)(x 3 + 2) 6. g(t) = (4t − 7)2 1 f (y) = y(2y − 1)(2y + 1) 8. f (x) = 4x 4 − 2 x 1 1 1 − 10. f (t) = g(x) = x +1 x −1 4 − t2 1 3 12. f (x) = h(x) = 2 2 x +x +1 1− x g(t) = (t 2 + 1)(t 3 + t 2 + 1) 3 14. f (x) = (2x 3 − 3)(17x 4 − 6x + 2) 1 1 − 2 15. g(z) = 2z 3z 2x 3 − 3x 2 + 4x − 5 16. f (x) = x2 2 17. g(y) = 2y(3y − 1)(y 2 + 2y + 3) x2 − 4 x2 + 4 t −1 g(t) = 2 t + 2t + 1 1 u(x) = (x + 2)2 1 v(t) = (t − 1)3 2x 3 + x 2 − 3x + 17 h(x) = 2x − 5 3x g(x) = 3 x + 7x − 5 1 f (t) = 1 2 t+ t 2 1 − 2 x g(x) = x 3 2 − 4 x3 x 1 x3 − 2 x +1 f (x) = 1 x4 + 2 x +1 3 h(x) = x − 6x 5 + 32 x −4 + 12 5 − 4x 2 + x 5 x3 31. y(x) = 3x − 1 4x 2 32. f (z) = 33. y(x) = x x +1 + x −1 3x 35. y(x) = x 3 − 4x + 5 x2 + 9 37. y(x) = 39. y(x) = 2x 2 x2 x +1 24. 25. 26. 27. 28. x(t) = 3 4 − 2 −5 t t 4 − 3)2 w + 10 w2 In Problems 41 through 50, write an equation of the line tangent to the curve y = f (x) at the given point P on the curve. Express the answer in the form ax + by = c. 41. y = x 3 ; P(2, 8) 42. y = 3x 2 − 4; P(1, −1) 1 ; P(2, 1) x −1 44. y = 2x − 45. y = x 3 + 3x 2 − 4x − 5; P(1, −5) 23. (t 2 40. h(w) = 19. 22. 1 + 2z + 2) 1 1 − 4t −2 3 2 3 36. w(z) = z 2z − 4 4z 38. z(t) = 4 3x − 4 5x z(z 2 34. u(t) = 43. y = 21. 2x − 3x 2 + 2x 4 5x 2 30. u(x) = 18. f (x) = 20. 128 29. y(x) = 46. y = 1 1 − 2 x x 1 ; P(0.5, −1) x −1 ; P(2, 4) 47. y = 4 3 − 3 ; P(−1, 7) x2 x 48. y = 3x − 2 ; P(2, 0.5) 3x + 2 49. y = 3x 2 ; P(−1, 3) x2 + x + 1 50. y = 6 ; P(2, −2) 1 − x2 51. Apply the formula in Example 5 to answer the following two questions. (a) If 1000 cm3 of water at 0◦ C is heated, does it initially expand or contract? (b) What is the rate (in cm3 /◦ C) at which it initially contracts or expands? 52. Susan’s weight in pounds is given by the formula W = (2 × 109 )/R 2 , where R is her distance in miles from the center of the earth. What is the rate of change of W with respect to R when R = 3960 mi? If Susan climbs a mountain, beginning at sea level, at what rate in ounces per (vertical) mile does her weight initially decrease? Basic Differentiation Rules SECTION 3.2 53. The conical tank shown in Fig. 3.2.5 has radius 160 cm and height 800 cm. Water is running out of a small hole in the bottom of the tank. When the height h of water in the tank is 600 cm, what is the rate of change of its volume V with respect to h? 129 62. (a) First write u 1 u 2 u 3 u 4 = (u 1 u 2 u 3 )u 4 to verify Eq. (16) for n = 4. (b) Then write u 1 u 2 u 3 u 4 u 5 = (u 1 u 2 u 3 u 4 )u 5 and apply the result in part (a) to verify Eq. (16) for n = 5. 63. Apply Eq. (16) to show that Dx ([ f (x)]n ) = n[ f (x)]n−1 · f (x) 120 if n is a positive integer and f (x) exists. 64. Use the result of Problem 63 to compute Dx [(x 2 + x +1)100 ]. 65. Use the result of Problem 63 to find g (x) given g(x) = (x 3 − 17x + 35)17 . 66. Find constants a, b, c, and d such that the graph of f (x) = ax 3 + bx 2 + cx + d 600 has horizontal tangent lines at the points (0, 1) and (1, 0). In connection with Problems 67 through 71, Figs. 3.2.8 through 3.2.11 show the curves y= FIGURE 3.2.5 The leaky tank of Problem 53. 54. Find the x- and y-intercepts of the straight line that is tangent to the curve y = x 3 + x 2 + x at the point (1, 3) (Fig. 3.2.6). 10 y= x3 + x2 2 0.4 4 (1, 3) 0 y= x3 −1 −4 −2 0 x 2 4 FIGURE 3.2.6 The tangent line of Problem 54. −2 −1 0 x 1 2 FIGURE 3.2.7 The tangent line of Problem 55. 55. Find an equation for the straight line that passes through the point (1, 5) and is tangent to the curve y = x 3 . [Suggestion: Denote by (a, a 3 ) the point of tangency, as indicated in Fig 3.2.7. Find by inspection small integral solutions of the resulting cubic equation in a.] 56. Find two lines through the point (2, 8) that are tangent to the curve y = x 3 . [See the suggestion for Problem 55.] 57. Prove that no straight line can be tangent to the curve y = x 2 at two different points. 0 − 0.8 −2 −2 (a, a3) 0 x −4 2 1 FIGURE 3.2.8 The graph 1 . of y = 1 + x2 −2 2 y= 1 x2/(1 + x2) y 0 −1 −1 −2 0 x 2 2 4 y = x3/(1 + x2) 1 y 0 −4 0 x FIGURE 3.2.9 The graph x . of y = 1 + x2 2 −2 y = x/(1 + x2) − 0.4 −1 y −10 y y 0 (1, 5) −4 0.8 y = 1/(1 + x2) 8 +x y 0 −20 for n = 0, 1, 2, and 3. 1 20 xn 1 + x2 −2 −2 4 −1 0 x 1 2 58. Find the two straight lines of slope −2 that are tangent to the curve y = 1/x. FIGURE 3.2.10 The graph x2 . of y = 1 + x2 59. Let n 2 be a fixed but unspecified integer. Find the x-intercept of the line that is tangent to the curve y = x n at the point P(x0 , y0 ). 67. Show that for n = 0 and n = 2, the curve has only a single point where the tangent line is horizontal (Figs. 3.2.8 and 3.2.10). 60. Prove that the curve y = x 5 + 2x has no horizontal tangents. What is the smallest slope that a line tangent to this curve can have? 68. When n = 1, there are two points on the curve where the tangent line is horizontal (Fig. 3.2.9). Find them. 61. Apply Eq. (16) with n = 3 and u 1 = u 2 = u 3 = f (x) to show that Dx ([ f (x)]3 ) = 3[ f (x)]2 · f (x). FIGURE 3.2.11 The graph x3 . of y = 1 + x2 69. Show that for n 3, (0, 0) is the only point on the graph of y= xn 1 + x2 at which the tangent line is horizontal (Fig. 3.2.11). 129 130 CHAPTER 3 The Derivative 70. Figure 3.2.12 shows the graph of the derivative f (x) of the function x3 f (x) = . 1 + x2 There appear to be two points on the graph of y = f (x) at which the tangent line has slope 1. Find them. 71. It appears in Fig. 3.2.12 that there are three points on the curve y = f (x) at which the tangent line is horizontal. Find them. 2 1001 V = V(T) y = Dxx3/(1 + x2) 1 V y 1000 (Tm , Vm) 0 −1 −4 −2 0 x 2 999 0 4 FIGURE 3.2.12 The graph x3 of y = Dx of 1 + x2 Problems 70 and 71. 5 V = 1000 V = Vm 10 T 15 Evidently a minimal volume Vm = V (Tm ) occurs at a critical temperature Tm ≈ 4 (◦ C). Given that the tangent line to the graph of V is horizontal at the point (Tm , Vm ), find: (a) the numerical values of Tm and Vm , and (b) the temperature T1 ≈ 8 (◦ C) at which the volume of the sample is again exactly 1000 cm3 . Comment: Because water that’s slightly warmer than the freezing point of 0◦ C is slightly denser than water at 0◦ C, the warmer water sinks to the bottom as a cooling lake freezes. But ice is less dense, so it floats on the surface. Consequently, ice at the surface traps somewhat warmer water at the bottom of the lake—which otherwise might freeze solid. This phenomenon is responsible for the survival and evolution of life forms that can withstand cold water but not freezing. In Problems 73 through 78, sketch the graph of the given function f and determine where it is differentiable. Recall the definition of one-sided derivatives in Problem 54 of Section 3.1, as well as the fact that f (a) exists if and only if f − (a) = f + (a). 20 FIGURE 3.2.13 The temperature-volume graph of Problem 72. 72. Much of life on earth (as we know it) depends critically on the variation of water density with temperature. Consider a sample of water than has a volume of exactly 1000 cm3 when measured at precisely 0◦ C. Figure 3.2.13 shows a graph of its volume function V (T ) as given by the formula in Example 5. The surprise is that, as the temperature is increased, the sample initially contracts rather than expands in volume. 73. f (x) = |x 3 | 74. f (x) = x 3 + |x 3 | 2 + 3x 2 if x < 1, 75. f (x) = 3 + 2x 3 if x 1 ⎧ ⎨x 4 if x < 1, 1 76. f (x) = ⎩2 − 4 if x 1 x ⎧ ⎨ 1 if x < 1, 77. f (x) = 2 − x ⎩x if x 1 ⎧ ⎨ 12 if x < 3, 78. f (x) = (5 − x)2 ⎩ 2 x − 3x + 3 if x 3 3.3 THE CHAIN RULE We saw in Section 3.2 how to differentiate powers of the independent variable, but we often need to differentiate powers of rather general (or even unknown) functions. For instance, suppose that y = u3 (1) where u is in turn a function of x. Then the extended product rule [Eq. (16) in Section 3.2] yields dy = Dx u 3 = Dx (u · u · u) = u · u · u + u · u · u + u · u · u dx where u = du/d x. After we collect terms, we find that du dy = 3u 2 u = 3u 2 . dx dx (2) Is it a surprise that the derivative of u 3 is not simply 3u 2 , which you might expect in analogy with the correct formula Dx x 3 = 3x 2 ? There is an additional factor du/d x, whose presence may seem more natural if we differentiate y in Eq. (1) with respect to u, and write dy = 3u 2 . du 130 The Chain Rule SECTION 3.3 131 Then the derivative formula in (2) takes the form dy dy du = · . dx du d x (3) Equation (3), the chain rule, holds for any two differentiable functions y = f (u) and u = g(x). The formula in Eq. (2) is simply the special case of (3) with f (u) = u 3 . EXAMPLE 1 If y = (3x 2 + 5)17 , it would be impractical to write the binomial expansion of the seventeenth power of 3x 2 + 5 before differentiating. The Expand command in a typical computer algebra system yields a polynomial in x having 18 terms, some of which have 15-digit coefficients: (3x 2 + 5)17 = 762939453125 + 7781982421875x 2 + · · · + 186911613281250x 18 + · · · + 129140163x 34 . (Each ellipsis replaces seven omitted terms.) But if we simply write y = u 17 with u = 3x 2 + 5, then dy = 17u 16 du and du = 6x. dx Hence the chain rule yields dy du dy = · = 17u 16 · 6x dx du d x = 17(3x 2 + 5)16 · 6x = 102x(3x 2 + 5)16 . ◗ The formula in (3) is one that, once learned, is unlikely to be forgotten. Although dy/du and du/d x are not fractions—they are merely symbols representing the derivatives f (u) and g (x)—it is much as though they were fractions, with the du in the first factor canceling the du in the second factor: dy du dy dy du · = · = . [Invalid cancellation!] du d x dx du d x x barrels crude oil Process 1 u liters gasoline Process 2 y grams petrochemical FIGURE 3.3.1 The two-process oil refinery (Example 2). But you should realize that such “cancellation” no more proves the chain rule than canceling two copies of the symbol d proves that dy y dy = = . [An absurdity!] dx x dx It is nevertheless an excellent way to remember the chain rule. Such manipulations with differentials are so suggestive (even when invalid) that they played a substantial role in the early development of calculus in the seventeenth and eighteenth centuries. Many formulas were thereby produced (and later proved valid), as were some formulas that were incorrect. EXAMPLE 2 For a physical interpretation of the chain rule, imagine an oil refinery that first makes u liters of gasoline from x barrels of crude oil. Then, in a second process, the refinery makes y grams of a marketable petrochemical from the u liters of gasoline. (The two processes are illustrated in Fig. 3.3.1.) Then y is a function of u 131 132 CHAPTER 3 The Derivative and u is a function of x, so the final output y is a function also of the input x. Consider the units in which the derivatives of these functions are measured. dy g : (grams of petrochemical du L per liter of gasoline) L du : d x barrel dy g : d x barrel (liters of gasoline per barrel of oil) (grams of petrochemical per barrel of oil) When we include the units in the chain rule equation dy dy du = · , dx du d x we get dy g = d x barrel dy g du L / dy du g · = · . du L / d x barrel du d x barrel The handy cancellation of units seems to confirm the validity of the chain rule (at least in this application). For example, if we get 3 g of petrochemical per liter of gasoline and 75 L of gasoline per barrel of oil, how could we fail to get 225 = 3 · 75 g of ◗ petrochemical per barrel of oil? The Chain Rule in Function Notation Although Eq. (3) is a memorable statement of the chain rule in differential notation, it has the disadvantage of not specifying the values of the variables at which the derivatives are evaluated. This problem can be solved by the use of function notation for the derivatives. Let us write y = f (u), u = g(x) y = h(x) = f (g(x)). Then du = g (x), dx dy = h (x), dx and dy = f (u) = f (g(x)). du Substituting these derivatives into the chain rule formula dy dy du = · dx du d x (3) h (x) = f (g(x)) · g (x). (4) recasts it in the form This version of the chain rule gives the derivative of the composition h = f ◦ g of two functions f and g in terms of their derivatives. THEOREM 1 The Chain Rule Suppose that g is differentiable at x and that f is differentiable at g(x). Then the composition h = f ◦ g defined by h(x) = f (g(x)) is differentiable at x, and its derivative is h (x) = f (g(x)) · g (x). 132 (4) The Chain Rule SECTION 3.3 133 REMARK The chain rule in (4) shows that the derivative of the composition h = f ◦ g is a product of the derivatives of f and g. Note, however, that these two derivatives are evaluated at different points. The derivative g of the inner function is evaluated at x, whereas the derivative f of the outer function is evaluated at g(x) (rather than at the same point x). EXAMPLE 3 In Example 1 we applied the differential form of the chain rule in (3) to differentiate the function h(x) = (3x 2 + 5)17 . To apply the functional form of the chain rule in (4), we must first identify the outer function f (x) = x 17 , for which f (x) = 17x 16 , and the inner function g(x) = 3x 2 + 5, for which g (x) = 6x. Then h (x) = f (g(x)) · g (x) = f (3x 2 + 5) · (3x 2 + 5) = 17(3x 2 + 5)16 · 6x = 102x(3x 2 + 5)16 . ◗ The Proof of the Chain Rule To outline a proof of the chain rule, suppose that we are given differentiable functions y = f (u) and u = g(x) and want to compute the derivative y f (g(x + x)) − f (g(x)) dy = lim = lim . x→0 x x→0 dx x (5) The differential form of the chain rule suggests the factorization y y u = x u x (6) where u = g(x + x) − g(x) and y = f (u + u) − f (u). For x fixed, the factorization in Eq. (6) is valid if g (x) = 0, because g (x) = u du = lim = 0 x→0 x dx implies that u = 0 if x = 0 is sufficiently small—for if so, then u = (u/x) · x is the product of nonzero numbers. But the fact that g is differentiable, and therefore continuous, at the point x (see Theorem 2 in Section 3.4) implies that u = g(x + x) − g(x) → 0 as x → 0. The product law of limits therefore gives y u dy du y u dy = lim · = lim · lim = · . x→0 u x u→0 u x→0 x dx du d x Thus we have shown that Dx [ f (g(x))] = f (g(x)) · g (x) at any point x at which g (x) = 0. But if g (x) = 0, then it is entirely possible that u is zero for some or all nonzero values of x approaching zero—in which case the factorization in (6) is invalid. Our proof of the chain rule is therefore incomplete. In Section 4.2 we give a proof that does not require the assumption that g (x) = 0. 133 134 CHAPTER 3 The Derivative The Generalized Power Rule If we substitute g(x) = u and g (x) = du/d x into Eq. (4) with h (x) = Dx f (g(x)) = Dx f (u), we get the hybrid form Dx [ f (u)] = f (u) · du dx (7) of the chain rule that frequently is the most useful form for purely computational purposes. Recall that the subscript x in Dx specifies that f (u) is being differentiated with respect to x, not with respect to u. Let us set f (u) = u n in Eq. (7), where n is an integer. Because f (u) = nu n−1 , we thereby obtain Dx u n = nu n−1 du , dx (8) the chain rule version of the power rule. Since u = g(x) is a differentiable function, Eq. (8) implies that Dx [g(x)]n = n[g(x)]n−1 · Dx [g(x)]. (9) [If n − 1 < 0, we must add the proviso that g(x) = 0 in order for the right-hand side in Eq. (9) to be meaningful.] We refer to this chain rule version of the power rule as the generalized power rule. REMARK We may interpret the operator form in (9) as describing a chain rule procedure in which we work from the outside to the inside—differentiating first the outer function and then the inner function. This outside-inside procedure is illustrated in the next example. EXAMPLE 4 To differentiate y= (2x 3 1 , − x + 7)2 we first write y = (2x 3 − x + 7)−2 in order to apply the generalized power rule, Eq. (9), with n = −2. This gives dy = (−2)(2x 3 − x + 7)−3 · Dx (2x 3 − x + 7) dx derivative of outer function 2(1 − 6x 2 ) = (−2)(2x 3 − x + 7)−3 · (6x 2 − 1) = . (2x 3 − x + 7)3 ◗ derivative of inner function EXAMPLE 5 Find the derivative of the function z−1 5 h(z) = . z+1 Solution The key to applying the generalized power rule is observing what the given function is a power of. Here, h(z) = u 5 , 134 where u = z−1 , z+1 The Chain Rule SECTION 3.3 135 and z, not x, is the independent variable. Hence we apply first Eq. (8) and then the quotient rule to get du z−1 4 z−1 Dz =5 dz z+1 z+1 z − 1 4 (1)(z + 1) − (z − 1)(1) =5 · z+1 (z + 1)2 z−1 4 2 10(z − 1)4 =5 · = . z+1 (z + 1)2 (z + 1)6 h (z) = 5u 4 ◗ The importance of the chain rule goes far beyond the power function differentiations illustrated in Examples 1, 4, and 5. We shall learn in later sections how to differentiate exponential, logarithmic, and trigonometric functions. Each time we learn a new differentiation formula—for the derivative f (x) of a new function f (x)—the formula in Eq. (7) immediately provides us with the chain rule version of that formula, Dx f (u) = f (u)Dx u. The step from the power rule Dx x n = nx n−1 to the generalized power rule Dx u n = nu n−1 Dx u is our first instance of this general phenomenon. Rate-Of-Change Applications Suppose that the physical or geometric quantity p depends on the quantity q, which in turn depends on time t. Then the dependent variable p is a function both of the intermediate variable q and of the independent variable t. Hence the derivatives that appear in the chain rule formula d p dq dp = dt dq dt are rates of change (as in Section 3.1) of these variables with respect to one another. For instance, suppose that a spherical balloon is being inflated or deflated. Then its volume V and its radius r are changing with time t, and dV d V dr = . dt dr dt r Remember that a positive derivative signals an increasing quantity and that a negative derivative signals a decreasing quantity. FIGURE 3.3.2 The spherical balloon with volume V = 43 πr 3 . EXAMPLE 6 A spherical balloon is being inflated (Fig. 3.3.2). The radius r of the balloon is increasing at the rate of 0.2 cm/s when r = 5 cm. At what rate is the volume V of the balloon increasing at that instant? Solution Given dr/dt = 0.2 cm/s when r = 5 cm, we want to find d V /dt at that instant. Because the volume of the balloon is V = 43 πr 3 , we see that d V /dr = 4πr 2 . So the chain rule gives dV dr d V dr = · = 4πr 2 = 4π(5)2 (0.2) ≈ 62.83 dt dr dt dt at the instant when r = 5 cm. (cm3 /s) ◗ 135 136 CHAPTER 3 The Derivative In Example 6 we did not need to know r explicitly as a function of t. But suppose we are told that after t seconds the radius (in centimeters) of an inflating balloon is r = 3 + (0.2)t (until the balloon bursts). Then the volume of this balloon is t 3 4 3 4 , V = πr = π 3 + 3 3 5 so d V /dt is given explicitly as a function of t by 4 t 2 1 4 t 2 dV = π(3) 3 + = π 3+ . dt 3 5 5 5 5 EXAMPLE 7 Imagine a spherical raindrop that is falling through water vapor in the air. Suppose that the vapor adheres to the surface of the raindrop in such a way that the time rate of increase of the mass M of the droplet is proportional to the surface area S of the droplet. If the initial radius of the droplet is, in effect, zero and the radius is r = 1 mm after 20 s, when is the radius 3 mm? Solution We are given dM = k S, dt where k is some constant that depends upon atmospheric conditions. Now M = 43 πρr 3 and (10) S = 4πr 2 , where ρ denotes the density of water. Substitution of the chain rule results in d d M dr dM = · = dt dr dt 4 πρr 3 3 dr · dr dr = 4πρr 2 · dt dt and k S = k · 4πr 2 into Eq. (10) then yields 4πρr 2 so it follows that dr = 4πkr 2 , dt dr k = , dt ρ a constant. So the radius of the droplet grows at a constant rate. Thus if it takes 20 s ◗ for r to grow to 1 mm, it will take 1 min for r to grow to 3 mm. 3.3 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The chain rule can be expressed in the form 2. 3. 4. 5. 6. 7. 136 du dv du = · . dt dv dt The chain rule can be expressed in the form Dx [ f (g(x))] = f (g(x)) · g (x). The generalized power rule states that Dx [ f (x)]m = m[ f (x)]m−1 · f (x) if m is an integer and the right-hand side in the last equation is defined. According to the generalized power rule, Dx (3x + 5)17 = 51(3x + 5)16 . If h = f ◦ g, then h (x) = f (g(x)) · g (x). dy = (−2)(2x 3 − x + 7)−3 · (6x − 1). If y = y(x) = (2x 3 − x + 7)−2 , then dx If x −1 4 x −1 x −1 5 , then h (x) = 5 · Dx . h(x) = x +1 x +1 x +1 The Chain Rule SECTION 3.3 137 8. Given: Dx (sin x) = cos x. Then Dx (sin x)5 = 5(sin x)4 cos x. 9. Given: Dx (sin x) = cos x. Then Dx [sin(x 5 )] = 5x 4 cos(x 5 ). du = 7(3x 2 + 2x)6 . 10. If u = u(x) = (x 3 + x 2 )7 then dx 3.3 CONCEPTS: QUESTIONS AND DISCUSSION 1. As a mathematical quiz show contestant, you are asked to calculate the value F (7) for the composition F = f ◦ g. The functions f and g are unknown, but you are permitted to ask exactly three questions regarding numerical values of these functions and/or their derivatives at specified points. What three questions should you ask? 2. Write the function-notation form of the chain rule formula in Problem 63 for a composition F = f ◦ g ◦ h of three functions. What numerical data are now needed to calculate the numerical value F (7)? 3.3 PROBLEMS Find dy/d x in Problems 1 through 12. 1. y = (3x + 4)5 1 3. y = 3x − 2 5. y = (x 2 + 3x + 4)3 7. y = (2 − x)4 (3 + x)7 x +2 9. y = (3x − 4)3 11. y = [1 + (1 + x)3 ]4 2. y = (2 − 5x)3 1 4. y = (2x + 1)3 6. y = (7 − 2x 3 )−4 8. y = (x + x 2 )5 (1 + x 3 )2 (1 − x 2 )3 10. y = (4 + 5x + 6x 2 )2 12. y = [x + (x + x 2 )−3 ]−5 In Problems 13 through 20, express the derivative dy/d x in terms of x without first rewriting y as a function of x. 13. y = (u + 1)3 and u= 1 x2 1 1 − 2 and u = 2x + 1 2u 3u 15. y = (1 + u 2 )3 and u = (4x − 1)2 1 16. y = u 5 and u = 3x − 2 1 17. y = u(1 − u)3 and u = 4 x u x 18. y = and u = u+1 x +1 1 2 4 3 19. y = u (u − u ) and u = 2 x u 2 20. y = and u = x − 4 (2u + 1) x 14. y = 31. f (u) = (1 + u)3 (1 + u 2 )4 32. g(w) = (w2 − 3w + 4)(w + 4)5 −2 1 −1 33. h(v) = v − 1 − v 1 1 −4 1 34. p(t) = + 2 + 3 t t t 1 35. F(z) = (3 − 4z + 5z 5 )10 36. G(x) = {1 + [x + (x 2 + x 3 )4 ]5 }6 39. y = (x 2 − 1)2 = x 4 − 2x 2 + 1 40. y = (1 − x)3 = 1 − 3x + 3x 2 − x 3 1 2 + 5x 3 21. f (x) = (2x − x 2 )3 22. f (x) = 1 (1 − x 2 )4 x +1 7 25. f (x) = x −1 24. f (x) = (x 2 − 4x + 1)3 26. f (x) = 28. h(z) = z 2 (z 2 + 4)3 In Problems 37 through 44, dy/d x can be found in two ways— one way using the chain rule, the other way without using it. Use both techniques to find dy/d x and then compare the answers. (They should agree!) −1 1 3 4 12 37. y = (x ) = x 38. y = x = x In Problems 21 through 26, identify a function u of x and an integer n = 1 such that f (x) = u n . Then compute f (x). 23. f (x) = 27. g(y) = y + (2y − 3)5 1 3 29. F(s) = s − 2 s 1 2 30. G(t) = t 2 + 1 + t (x 2 + x + 1)4 (x + 1)4 Differentiate the functions given in Problems 27 through 36. 41. y = (x + 1)4 = x 4 + 4x 3 + 6x 2 + 4x + 1 1 42. y = (x + 1)−2 = 2 x + 2x + 1 1 43. y = (x 2 + 1)−1 = 2 x +1 44. y = (x 2 + 1)2 = (x 2 + 1)(x 2 + 1) We shall see in Section 3.7 that Dx [sin x] = cos x (provided that x is in radian measure). Use this fact and the chain rule to find the derivatives of the functions in Problems 45 through 48. 45. f (x) = sin(x 3 ) 46. g(t) = (sin t)3 47. g(z) = (sin 2z)3 48. k(u) = sin(1 + sin u) 137 138 CHAPTER 3 The Derivative 49. A pebble dropped into a lake creates an expanding circular ripple (Fig. 3.3.3). Suppose that the radius of the circle is increasing at the rate of 2 in./s. At what rate is its area increasing when its radius is 10 in.? FIGURE 3.3.3 Expanding circular ripple in a lake (Problem 49). 50. The area of a circle is decreasing at the rate of 2π cm2 /s. At what rate is the radius of the circle decreasing when its area is 75π cm2 ? 51. Each edge x of a square is increasing at the rate of 2 in./s. At what rate is the area A of the square increasing when each edge is 10 in.? 52. Each edge of an equilateral triangle is increasing at 2 cm/s (Fig. 3.3.4). At what rate is the area of the triangle increasing when each edge is 10 cm? x 55. Given: G(t) = f (h(t)), h(1) = 4, f (4) = 3, and h (1) = −6. Find G (1). 56. Suppose that f (0) = 0 and that f (0) = 1. Calculate the derivative of f ( f ( f (x))) at x = 0. 57. Air is being pumped into a spherical balloon in such a way that its radius r is increasing at the rate of dr/dt = 1 cm/s. What is the time rate of increase, in cubic centimeters per second, of the balloon’s volume when r = 10 cm? 58. Suppose that the air is being pumped into the balloon of Problem 57 at the constant rate of 200π cm3 /s. What is the time rate of increase of the radius r when r = 5 cm? 59. Air is escaping from a spherical balloon at the constant rate of 300π cm3 /s. What is the radius of the balloon when its radius is decreasing at the rate of 3 cm/s? 60. A spherical hailstone is losing mass by melting uniformly over its surface as it falls. At a certain time, its radius is 2 cm and its volume is decreasing at the rate of 0.1 cm3 /s. How fast is its radius decreasing at that time? 61. A spherical snowball is melting in such a way that the rate of decrease of its volume is proportional to its surface area. At 10 A . M . its volume is 500 in.3 and at 11 A . M . its volume is 250 in.3 . When does the snowball finish melting? (See Example 7.) 62. A cubical block of ice with edges 20 in. long begins to melt at 8 A . M . Each edge decreases at a constant rate thereafter and each is 8 in. long at 4 P. M . What was the rate of change of the block’s volume at noon? 63. Suppose that u is a function of v, that v is a function of w, that w is a function of x, and that all these functions are differentiable. Explain why it follows from the chain rule that du du dv dw = · · . dx dv dw d x x h=? 1 x 2 1 x 2 FIGURE 3.3.4 The triangle of Problem 52 with area A = 12 xh. 53. A cubical block of ice is melting in such a way that each edge decreases steadily by 2 in. every hour. At what rate is its volume decreasing when each edge is 10 in. long? 54. Find f (−1), given f (y) = h(g(y)), h(2) = 55, g(−1) = 2, h (2) = −1, and g (−1) = 7. 64. Let f be a differentiable function such that f (1) = 1. If F(x) = f (x n ) and G(x) = [ f (x)]n (where n is a fixed integer), show that F(1) = G(1) and that F (1) = G (1). Recall from Example 13 in Section 2.2 that Dx √ 1 x = √ . 2 x Use (only) this fact and the chain rule to calculate the derivative of each function given in Problems 65 through 68. √ 66. h(x) = x 3/2 65. h(x) = x + 4 √ 2 3/2 67. h(x) = (x + 4) 68. h(x) = |x| = x 2 3.4 DERIVATIVES OF ALGEBRAIC FUNCTIONS We saw in Section 3.3 that the chain rule yields the differentiation formula du (1) dx if u = f (x) is a differentiable function and the exponent n is an integer. We shall see in Theorem 1 of this section that this generalized power rule holds not only when the exponent is an integer, but also when it is a rational number r = p/q (where p and q are integers and q = 0). Recall that rational powers are defined in terms of integral roots and powers as follows: √ √ p q u p/q = u p = q u . Dx u n = nu n−1 138 Derivatives of Algebraic Functions SECTION 3.4 139 We consider first the case of a rational power of the independent variable x: y = x p/q , (2) where p and q are integers with q positive. In Problems√72 through 75 we illustrate the proof that the derivative of the root function f (x) = q x is given by Dx x 1/q = 1 1 (1/q)−1 x = x −(q−1)/q q q (3) for x > 0; essentially the same proof works for x < 0 if q is odd (so that no even root of a negative number is involved). Thus the power rule—which we established in Section 3.2 only for integral exponents—also holds if the exponent of x is the reciprocal of a positive integer. Consequently, we can apply Eq. (1) with n = p and u = x 1/q to differentiate the rational power of x in (2): p Dx x p/q = Dx x 1/q = p x 1/q = p x 1/q = p−1 · Dx x 1/q p−1 1 (1/q)−1 · x q p ( p/q)−(1/q)+(1/q)−1 x ; q therefore p Dx x p/q = x ( p/q)−1 . q Thus we have shown that the power rule Dx x r = r x r −1 (4) holds if the exponent r = p/q is a rational number (subject to the conditions previously mentioned). Using Eq. (4) we can differentiate a simple “radical” (or “root”) function by first rewriting it as a power with a fractional exponent. EXAMPLE 1 √ 1 1 (a) Dx x = Dx x 1/2 = x −1/2 = √ . 2 2 x √ dy 3√ 3 x. = x 1/2 = (b) If y = x 3 , then dx 2 2 1 2 2 = t −2/3 , then g (t) = − t −5/3 = − √ . (c) If g(t) = √ 3 2 3 5 3 t 3 t 2 y' = y 1 2 x 1 0 y= √ REMARK In parts (a) and (b) of Example 1 it is necessary that x 0 in order that x be defined. In part (a) it is, moreover, necessary that x = 0; if x = 0 then the formula x Dx 0 0.5 1 x 1.5 FIGURE 3.4.1 The graphs of √ 1 f (x) = x and f (x) = √ . 2 x ◗ 2 √ 1 x = √ 2 x would involve division by zero. Figure the function f (x) = √ √ 3.4.1 shows the graphs of x and its derivative f (x) = 1/(2 x) for x > 0. Note that f (x) → ∞ as x → 0+ , √ further emphasizing the fact that f (x) = x is not differentiable at x = 0. 139 140 CHAPTER 3 The Derivative The Generalized Power Rule For the more general form of the power rule, let y = ur where u is a differentiable function of x and r = p/q is rational. Then dy = r u r −1 du by Eq. (4), so the chain rule gives dy du dy du = · = r u r −1 . dx du d x dx Thus Dx u r = r u r −1 du , dx (5) which is the generalized power rule for rational exponents. THEOREM 1 Generalized Power Rule If r is a rational number, then Dx [ f (x)]r = r [ f (x)]r −1 · f (x) (6) wherever the function f is differentiable and the right-hand side is defined. For the right-hand side in Eq. (6) to be “defined” means that f (x) exists, there is no division by zero, and no even root of a negative number appears. 3 y = 4 − x2 2 EXAMPLE 2 Dx 4 − x 2 = Dx (4 − x 2 )1/2 = 12 (4 − x 2 )−1/2 · Dx (4 − x 2 ) 1 y 0 −1 −2 −3 −3 = 12 (4 − x 2 )−1/2 · (−2x); y' = − −2 x 4 − x2 −1 Dx 0 x 1 2 FIGURE√3.4.2 The graphs of f (x) = 4 − x 2 and −x . f (x) = √ 4 − x2 3 x 4 − x2 = −√ 4 − x2 (7) except where x = ±2 (division by zero) or where |x| > 2 (square root of a negative number). Thus Eq. (7) holds if −2 < x < 2. In writing derivatives of algebraic functions, we ordinarily omit such disclaimers unless they√are pertinent to some specific purpose at hand. But note in Fig. 3.4.2 that if f (x) = 4 − x 2 then f (x) → +∞ as ◗ x → −2+ and f (x) → −∞ as x → +2− . A template for the application of the generalized power rule is Dx ([∗ ∗ ∗]n ) = n[∗ ∗ ∗]n−1 Dx [∗ ∗ ∗], where ∗ ∗ ∗ represents a function of x and (as we now know) n can be either an integer or a fraction (a quotient of integers). But to differentiate a power of a function, we must first recognize what function it is a power of. So to differentiate a function involving roots (or radicals), we first “prepare” it for an application of the generalized power rule by rewriting it as a power function with fractional exponent. Examples 3, 5, and 6 illustrate this technique. √ 2 , then EXAMPLE 3 If y = 5 x 3 − √ 3 x y = 5x 3/2 − 2x −1/3 , 140 Derivatives of Algebraic Functions SECTION 3.4 141 so dy =5· dx 3 1/2 2 1 15 1/2 2 −4/3 15 √ x x + x = x+ √ . − 2 · − x −4/3 = 3 4 2 3 2 3 2 3 x ◗ EXAMPLE 4 With f (x) = 3 − 5x and r = 7, the generalized power rule yields Dx [(3 − 5x)7 ] = 7(3 − 5x)6 Dx (3 − 5x) = 7(3 − 5x)6 (−5) = −35(3 − 5x)6 . ◗ EXAMPLE 5 With f (x) = 2x 2 − 3x + 5 and r = 12 , the generalized power rule yields Dx 2x 2 − 3x + 5 = Dx (2x 2 − 3x + 5)1/2 1 = (2x 2 − 3x + 5)−1/2 Dx (2x 2 − 3x + 5) 2 4x − 3 . = √ ◗ 2 2x 2 − 3x + 5 EXAMPLE 6 If 10 3 x = 5t + (3t − 1)4 then Eq. (5) with u = 5t + (3t − 1)4/3 and with independent variable t gives du dx = 10u 9 · dt dt = 10 5t + (3t = 10 5t + (3t = 10 5t + (3t dx = 10 5t + (3t dt · Dt 5t + (3t − 1)4/3 9 − 1)4/3 · Dt (5t) + Dt (3t − 1)4/3 9 − 1)4/3 · 5 + 43 (3t − 1)1/3 · 3 ; 9 − 1)4/3 · 5 + 4(3t − 1)1/3 . − 1)4/3 9 ◗ Example 6 illustrates the fact that we apply the chain rule (or generalized power rule) by working from the outside to the inside. At each step the derivative of the outside function is multiplied by the derivative of the inside function. We continue until no “inside function” remains undifferentiated. Does the process remind you of peeling an onion, one layer at a time, until its core is reached? Differentiability and Vertical Tangent Lines Whereas polynomials and rational functions are both continuous and differentiable wherever they are defined, simple algebraic functions can be continuous at points where their derivatives do not exist. y EXAMPLE 7 If f (x) = |x| = f (x) = |x | x FIGURE 3.4.3 The graph of f (x) = |x|. √ x2 denotes the absolute value function, then for x = 0 we find that 2 1/2 1 2 −1/2 x x −1 = = 2 (x ) (2x) = √ = f (x) = Dx (x ) 2 +1 |x| x if x < 0, if x > 0. Thus f is differentiable at every point except possibly for the origin x = 0. In fact, the graph of f (x) = |x| in Fig. 3.4.3 makes it clear that the difference quotient |x| f (x) − f (0) = x −0 x 141 142 CHAPTER 3 The Derivative has left-hand limit −1 and right-hand limit +1 at x = 0. Thus the absolute value function is not differentiable at the isolated point x = 0, where the graph y = |x| has a “corner point” rather than a tangent line. (Can you think of a continuous function ◗ whose graph has infinitely many such corner points?) EXAMPLE 8 Figure 3.4.4 shows the graph of the cube-root function √ y = 3 x = x 1/3 , y y = x1/3 and illustrates another way in which a function can fail to be differentiable at an isolated point. Its derivative, x 1 1 dy , = x −2/3 = √ 3 2 dx 3 3 x increases without bound as x → 0 but does not exist at x = 0. Therefore, the definition of tangent line does not apply to this graph at (0, 0). Nevertheless, from the figure it seems appropriate to regard the vertical line x = 0 as the line tangent to the curve ◗ y = x 1/3 at the point (0, 0). FIGURE 3.4.4 The graph of the cube root function. DEFINITION Vertical Tangent Line The curve y = f (x) has a vertical tangent line at the point (a, f (a)) provided that f is continuous at a and f (x) → +∞ as x → a. (8) y ( 12 , 12 ) y = x 1 − x2 (−1, 0) (1, 0) x (− Thus the graph of the continuous function f (x) = x 1/3 of Example 8 has a vertical tangent line at the origin, even though f is not differentiable at x = 0. Note that the requirement that f be continuous at x = a implies that f (a) must be defined. Thus it would be pointless to ask about a line (vertical or not) tangent to the curve y = 1/x where x = 0. If f is defined (and differentiable) on only one side of x = a, we mean in Eq. (8) that | f (x)| → +∞ as x approaches a from that side. EXAMPLE 9 Find the points on the curve y = f (x) = x 1 − x 2 , −1 x 1, 1 , 1 − ) 2 2 at which the tangent line is either horizontal or vertical. FIGURE 3.4.5 The graph of √ f (x) = x 1 − x 2 , −1 x 1 (Example 9). Solution We differentiate using first the product rule and then the chain rule: x (1 − x 2 )−1/2 (−2x) 2 1 − 2x 2 . = (1 − x 2 )−1/2 [(1 − x 2 ) − x 2 ] = √ 1 − x2 f (x) = (1 − x 2 )1/2 + 1.5 (0, 1) 1 y 0.5 0 −0.5 −2 −1 0 x 1 FIGURE √ 3.4.6 The graph of 5 y = 1 − x 2 with a cusp at (0, 1). 142 2 √ only when the numerator 1 − 2x 2 is zero—that is, when x = ±1/ 2. Now f (x) = 0 √ Because f√ (±1/ 2) = ±1/2,√the curve has a horizontal tangent line at each of the two points (1/ 2, 1/2) and (−1/ 2, −1/2). √ We also observe that the denominator 1 − x 2 approaches zero as x → −1+ and as x → +1− . Because f (±1) = 0, we see that the curve has a vertical tangent line at each of the two points (1, 0) and (−1, 0). The graph of f is shown in ◗ Fig. 3.4.5. √ 5 EXAMPLE 10 Figure 3.4.6 shows the graph of the function f (x) = 1 − x 2 , which appears to have a sharp “cusp” (rather than a corner) at the point (0, 1). Because the absolute value of the derivative f (x) = − 25 x −3/5 approaches +∞ as x → 0, the curve ◗ y = f (x) has a vertical tangent at that point. Derivatives of Algebraic Functions SECTION 3.4 143 Whereas the preceding examples show that a function can be continuous without being differentiable, the following theorem says that a function is continuous wherever it is differentiable. Thus differentiability of a function is a stronger condition than continuity alone. THEOREM 2 Differentiability Implies Continuity Suppose that the function f is defined in a neighborhood of a. If f is differentiable at a, then f is continuous at a. Proof Because f (a) exists, the product law for limits yields f (x) − f (a) lim [ f (x) − f (a)] = lim (x − a) · x→a x→a x −a f (x) − f (a) = lim (x − a) lim x→a x→a x −a = 0 · f (a) = 0. Thus lim f (x) = f (a), so f is continuous at a. x→a ◆ 3.4 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. √ 1. If f (x) = x, then f (x) = 12 x −1/2 . 2. Dx [x −3/2 ] = − 32 x −1/2 . 3. Suppose that r is a rational number and that f is a differentiable function of x. Then Dx [ f (x)]r = r [ f (x)]r −1 · f (x). √ 4. Dx 4 − x 2 = 12 (4 − x 2 )−1/2 . 5. If f is continuous at x = a and | f (x)| → +∞ as x → a, then the graph of f has a vertical tangent line at the point (a, f (a)). √ 6. If f (x) = x 1 − x 2 , then the graph of f has vertical tangent lines at the two points (1, 0) and (−1, 0). 7. If f is continuous at x = a then f (a) exists. 8. If f (a) exists then f is continuous at x = a. 9. If g(x) = |x − 1|+2 then g is continuous everywhere but fails to be differentiable at infinitely many points. 1 10. If h(x) = then the graph of h has a vertical tangent line at (0, 0). x 3.4 CONCEPTS: QUESTIONS AND DISCUSSION 1. (a) Can you define a function that is continuous everywhere and has a “corner point” at each integer point x = n, but is differentiable at every other point of the real line? (b) Can you define a function that is continuous everywhere and has a vertical tangent line at each integer point x = n, but is differentiable at every other point of the real line? 2. Suppose that the function f has the following property: Every point x of the real line lies in some closed interval [a, b] on which the graph of f is a semicircle having this interval as a diameter. Sketch a typical graph of such a function. Discuss the continuity and differentiability of f . Remark: The set of all endpoints of the closed intervals mentioned might (or might not) be the set of all integer points on the real line. 143 144 CHAPTER 3 The Derivative 3. In Question 2, you may have assumed that each endpoint of the interval [a, b] lies on exactly two such semicircles, one to the right and one to the left. Can you think of a function g whose graph consists entirely of semicircles, but does not satisfy this “two semicircles” condition? If so, discuss the differentiability of g. Suggestion: The construction of g might (or might not) involve the set {1, 12 , 13 , 14 , 15 , . . . } of all reciprocals of positive integers. 4. Suppose that the function f is continuous everywhere. At how many points do you suspect that f can fail to be differentiable? What’s the worst such function you can think of? 3.4 PROBLEMS Differentiate the functions given in Problems 1 through 44. √ √ 2 3 3 2. g(t) = 9 t 4 − √ 1. f (x) = 4 x 5 + √ 3 x t √ 1 3. f (x) = 2x + 1 4. h(z) = √ 3 7 − 6z 2 7 + 2u − 3u 4 6−x 6. φ(u) = 5. f (x) = √ √ 3 x u2 7. f (x) = (2x + 3) 3/2 2 −3/2 9. f (x) = (3 − 2x ) √ 11. f (x) = x 3 + 1 √ 13. f (x) = 2x 2 + 1 √ 15. f (t) = 2t 3 17. f (x) = (2x 2 − x + 7)3/2 18. g(z) = (3z 2 − 4)97 1 19. g(x) = (x − 2x 3 )4/3 20. f (t) = [t 2 + (1 + t)4 ]5 √ 21. f (x) = x 1 − x 2 2x + 1 22. g(x) = x −1 t2 + 1 23. f (t) = t2 − 1 y + 1 17 24. h(y) = y−1 1 3 25. f (x) = x − x 26. g(z) = √ 27. 28. 29. 30. z2 1 + z2 √ v+1 f (v) = v 5/3 x h(x) = 1 + x2 √ f (x) = 3 1 − x 2 √ g(x) = x + x 31. f (x) = x(3 − 4x)1/2 144 8. g(x) = (3x + 4) t − (1 + t 2 )1/2 t2 33. f (x) = (1 − x 2 )(2x + 4)1/3 32. g(t) = 34. f (x) = (1 − x)1/2 (2 − x)1/3 1 2 2 35. g(t) = 1 + (3t + 1)1/2 t 36. f (x) = x(1 + 2x + 3x 2 )10 4/3 3 −2/3 10. f (y) = (4 − 3y ) 1 12. g(z) = 4 (z + 3)2 t 14. f (t) = √ 1 + t4 1 16. g(t) = 3t 5 37. f (x) = 2x − 1 (3x + 4)5 38. h(z) = (z − 1)4 (z + 1)6 39. f (x) = (2x + 1)1/2 (3x + 4)1/3 40. f (x) = (1 − 3x 4 )5 (4 − x)1/3 √ √ 1+y+ 1−y 41. h(y) = 3 y5 √ 42. f (x) = 1 − 3 x √ 43. g(t) = t + t + t 1 3 44. f (x) = x 1 − 2 x +1 For each curve given in Problems 45 through 50, find all points on the graph where the tangent line is either horizontal or vertical. √ 45. y = x 2/3 46. y = x 4 − x 2 47. y = x 1/2 − x 3/2 49. y = √ x 1− x2 48. y = √ 50. y = 1 9 − x2 (1 − x 2 )(4 − x 2 ) In Problems 51 through 56, first write an equation of the line tangent to the given curve y = f (x) at the indicated point P. Then illustrate your result with a graphing calculator or computer by graphing both the curve and the tangent line on the same screen. √ 51. y = 2 x, at the point P where x = 4 √ 52. y = 3 3 x, at the point P where x = 8 √ 3 53. y = 3 x 2 , at the point P where x = −1 √ 54. y = 2 1 − x, at the point P where x = 34 √ 55. y = x 4 − x, at the point P where x = 0 √ 56. y = (1 − x) x, at the point P where x = 4 Derivatives of Algebraic Functions SECTION 3.4 145 In Problems 57 through 62, match the given graph y = f (x) of a function with the graph y = f (x) of its derivative among those shown in Figs. 3.4.13(a) through 3.4.13(f). 57. Figure 3.4.7 58. Figure 3.4.8 2 2 x2 / 3 y= 1 1 y 0 y 0 −1 −1 −2 −2 −1 1 0 x 2 FIGURE 3.4.7 y = x 2/3 (Problem 57). −2 −2 y = 1 − x2/3 1 0 x y 0 −1 −1 −1 0 x 1 2 −2 −2 y 0 y 0 −1 −1 −1 0 x 1 2 −2 −2 2 1 y 0 −3 −2 2 2 1 2 1 y 0 −1 −1 0 x 1 2 FIGURE 3.4.13(e) 0 x 1 2 −2 −1 0 x 2 1 y=x 2−x −1 FIGURE 3.4.13(d) −1 1 y 0 1 3 −1 2 1 1 FIGURE 3.4.13(c) 60. Figure 3.4.10 2 2 −2 −2 FIGURE 3.4.8 y = x 1/3 (Problem 58). 59. Figure 3.4.9 −2 −2 y = x1 /3 2 −2 −2 −1 0 x FIGURE 3.4.13(f) 63. The period of oscillation P (in seconds) of a simple pendu√ lum of length L (in feet) is given by P = 2π L/g, where g = 32 ft/s2 . Find the rate of change of P with respect to L when P = 2. FIGURE 3.4.9 y = 1 − x 2/3 (Problem 59). √ FIGURE 3.4.10 y = x 2 − x (Problem 60). 64. Find the rate of change of the volume V = 43 πr 3 of a sphere of radius r with respect to its surface area A = 4πr 2 when r = 10. 61. Figure 3.4.11 62. Figure 3.4.12 65. Find the two points on the circle x 2 + y 2 = 1 at which the slope of the tangent line is −2 (Fig. 3.4.14). 1 3 y = x 4 − x2 2 y 1 y 0 y 0 x2 + y2 = 1 y = (1 − x2)2 x2/3 −1 −2 −3 −2 −1.5 −1 − 0.5 0 0.5 1 1.5 2 x FIGURE 3.4.11 y = √ x 4 − x 2 (Problem 61). −1 −2 x −1 0 x 1 2 FIGURE 3.4.12 y= √ (1 − x 2 )2 x 2/3 (Problem 62). 2 3 FIGURE 3.4.14 The two tangent lines of Problem 65. 2 1 1 y 0 y 0 −1 −1 −2 −3 −2 66. Find the two points on the circle x 2 + y 2 = 1 at which the slope of the tangent line is 3. −1 0 x FIGURE 3.4.13(a) 1 2 −2 −2 −1 0 x FIGURE 3.4.13(b) 1 2 67. Find a line through the point P(18, 0) that is normal to the tangent line to the parabola y = x 2 at some point Q(a, a 2 ) (see Fig. 3.4.15). (Suggestion: You will obtain a cubic equation in the unknown a. Find by inspection a small integral root r . The cubic polynomial is then the product of a −r and a quadratic polynomial; you can find the latter by division of a − r into the cubic.) 145 146 CHAPTER 3 The Derivative y 12 20 4 P y = x2 8 (0, 2.5) x O y 1 Q 0 P 0 0 y = x2 /3 2 y = x2 y 4 y 10 P 3 10 x x2 + y2 = a2 0 −4 −5 20 FIGURE 3.4.15 The tangent and normal of Problem 67. 0 5 10 −1 −2 x FIGURE 3.4.16 The three normal lines of Problem 68. 68. Find three distinct lines through the point P(3, 10) that are normal to the parabola y = x 2 (Fig. 3.4.16). (See the suggestion for Problem 67. This problem will require a certain amount of calculator-aided computation.) 0 x 2 FIGURE 3.4.17 The two normal lines of Problem 69. FIGURE 3.4.18 The circle, radius, and tangent line of Problem 70. the evaluation of this limit using the algebraic identity s q − t q = (s − t) (s q−1 + s q−2 t + · · · + st q−2 + t q−1 ) . (10) q terms For instance, with s = x and t = a 1/q this identity yields (with q = 2, 3, and 5) the formulas 1/q 69. Find two distinct lines through the point P(0, 52 ) that are normal to the curve y = x 2/3 (Fig. 3.4.17). x − a = x 1/2 − a 1/2 x 1/2 + a 1/2 , 70. Verify that the line tangent to the circle x 2 + y 2 = a 2 at the point P is perpendicular to the radius OP (Fig. 3.4.18). 71. Consider the cubic equation x 3 = 3x + 8. If we differentiate each side with respect to x, we obtain 3x 2 = 3, which has the two solutions x = 1 and x = −1. But neither of these is a solution of the original cubic equation. What went wrong? Why does differentiation of both sides of the cubic equation give an invalid result? The derivation of the generalized power rule Dx u r = r u r −1 · Dx u (for r = p/q, a rational number) provided in this section depends on the assumed differentiability of the qth root function f (x) = x 1/q . If a > 0 and q is a positive integer, then the derivative of f is given by x 1/q − a 1/q f (a) = lim (9) x→a x −a provided that this limit exists. Problems 72 through 75 illustrate x −a = x 1/3 −a 1/3 x 2/3 +x 1/3 1/3 a (11) +a 2/3 , (12) and x − a = x 1/5 − a 1/5 x 4/5 + x 3/5 a 1/5 + x 2/5 a 2/5 + x 1/5 a 3/5 + a 4/5 . 72. Substitute (11) in the denominator in (9) Dx x 1/2 = 12 x −1/2 for x > 0. 73. Substitute (12) in the denominator in (9) Dx x 1/3 = 13 x −2/3 for x > 0. 74. Substitute (13) in the denominator in (9) Dx x 1/5 = 15 x −4/5 for x > 0. 75. Finally, explain how Eq. (10) can be applied case to prove that 1 Dx x 1/q = x −(q−1)/q q (13) to show that to show that to show that in the general if x > 0 and q is a positive integer. 3.5 MAXIMA AND MINIMA OF FUNCTIONS ON CLOSED INTERVALS x $5/ft y $5/ft $5/ft y $1/ft x FIGURE 3.5.1 The animal pen. Wall In applications we often need to find the maximum (largest) or minimum (smallest) value that a specified quantity can attain. The animal pen problem posed in Section 1.1 is a simple yet typical example of an applied maximum-minimum problem. There we investigated the animal pen shown in Fig. 3.5.1, with the indicated dollar-per-foot cost figures for its four sides. We showed that if $180 is allocated for material to construct this pen, then its area A = f (x) is given as a function of its base length x by f (x) = 35 x(30 − x), 0 x 30. (1) Hence the question of the largest possible area of the animal pen is equivalent to the purely mathematical problem of finding the maximum value attained by the function f (x) = 35 x(30 − x) on the closed interval [0, 30]. 146 Maxima and Minima of Functions on Closed Intervals SECTION 3.5 147 DEFINITION Maximum and Minimum Values If c is in the closed interval [a, b], then f (c) is called the minimum value of f (x) on [a, b] if f (c) f (x) for all x in [a, b]. Similarly, if d is in [a, b], then f (d) is called the maximum value of f (x) on [a, b] if f (d) f (x) for all x in [a, b]. Thus if f (c) is the minimum value and f (d) the maximum value of f (x) on [a, b], then (2) f (c) f (x) f (d) for all x in [a, b], and hence f (x) attains no value smaller than f (c) or larger than f (d). In geometric terms, (c, f (c)) is a low point and (d, f (d)) is a high point on the curve y = f (x), a x b, as illustrated in Figs. 3.5.2 and 3.5.3. y y High point High point f (b) f(d) Low point Low point f(c) a f (a) c d FIGURE 3.5.2 f (c) is the minimum value and f (d) is the maximum value of f (x) on [a, b]. b x a b x FIGURE 3.5.3 Maximum and minimum values can occur at the endpoints of an interval. Here f (a) is the minimum value and f (b) is the maximum value of f (x) on [a, b]. Theorem 1 (proved in Appendix E) says that a continuous function f on a closed interval [a, b] attains a minimum value f (c) and a maximum value f (d), so the inequalities in (2) hold: The curve y = f (x) over [a, b] has both a lowest point and a highest point. THEOREM 1 Maximum and Minimum Value Property If the function f is continuous on the closed interval [a, b], then there exist numbers c and d in [a, b] such that f (c) is the minimum value, and f (d) the maximum value, of f on [a, b]. y In short, a continuous function defined on a closed and bounded interval attains both a minimum value and a maximum value at points of the interval. Hence we see it is the continuity of the function 2 1 3 x(30 − x) 5 on the closed interval [0, 30] that guarantees that the maximum value of f exists and is attained at some point of the interval [0, 30]. Suppose that the function f is defined on the interval I . Examples 1 and 2 show that if either f is not continuous or I is not closed, then f may fail to attain maximum and minimum values at points of I . Thus both hypotheses in Theorem 1 are necessary. f (x) = f(x) = 2x x FIGURE 3.5.4 The graph of the function of Example 1. EXAMPLE 1 Let the continuous function f (x) = 2x be defined only for 0 x < 1, so that its domain of definition is not a closed interval. From the graph shown in Fig. 3.5.4, it is clear that f attains its minimum value 0 at x = 0. But f (x) = 2x attains no maximum value at any point of [0, 1). The only possible candidate for a maximum ◗ value would be the value 2 at x = 1, but f (1) is not defined. 147 148 CHAPTER 3 The Derivative EXAMPLE 2 The function f defined on the closed interval [0, 1] with the formula ⎧ ⎨ 1 if 0 < x 1, f (x) = x ⎩ 1 if x = 0 y 1 x f (x) = 1 is not continuous on [0, 1] because limx→0+ (1/x) does not exist (Fig. 3.5.5). This function does attain its minimum value of 1 at x = 0 and also at x = 1. But it attains no maximum value on [0, 1] because 1/x can be made arbitrarily large by choosing x ◗ positive and very close to zero. (1, 1) x 1 Local Maxima and Minima FIGURE 3.5.5 The graph of the function of Example 2. y Local maximum x Local minimum FIGURE 3.5.6 Local extrema. For a variation on Example 2, the function g(x) = 1/x with domain the open interval (0, 1) attains neither a maximum nor a minimum there. Once we know that the continuous function f does attain minimum and maximum values on the closed interval [a, b], the remaining question is this: Exactly where are these values located? We solved the animal pen problem in Section 2.1 on the basis of the following assumption, motivated by geometry: The function f (x) = 35 x(30 − x) attains its maximum value on [0, 30] at an interior point of that interval, a point at which the tangent line is horizontal. Theorems 2 and 3 of this section provide a rigorous basis for the method we used there. We say that the value f (c) is a local maximum value of the function f if f (x) f (c) for all x sufficiently near c. More precisely, if this inequality holds for all x that are simultaneously in the domain of f and in some open interval containing c, then f (c) is a local maximum of f . Similarly, we say that the value f (c) is a local minimum value of f if f (x) f (c) for all x sufficiently near c. As Fig. 3.5.6 shows, a local maximum is a point such that no nearby points on the graph are higher, and a local minimum is one such that no nearby points on the graph are lower. A local extremum of f is a value of f that is either a local maximum or a local minimum. THEOREM 2 Local Maxima and Minima Suppose that f is differentiable at c and is defined on a open interval containing c. If f (c) is either a local maximum value or a local minimum value of f , then f (c) = 0. Thus a local extremum of a differentiable function on an open interval can occur only at a point where the derivative is zero and, therefore, where the line tangent to the graph is horizontal. Proof of Theorem 2 Suppose, for instance, that f (c) is a local maximum value of f . The assumption that f (c) exists means that the right-hand and left-hand limits lim h→0+ f (c + h) − f (c) h and lim h→0− f (c + h) − f (c) h both exist and are equal to f (c). If h > 0, then f (c + h) − f (c) 0, h because f (c) f (c + h) for all small positive values of h. Hence, by a one-sided version of the squeeze law for limits (in Section 2.3), this inequality is preserved when we take the limit as h → 0. We thus find that f (c) = lim h→0+ 148 f (c + h) − f (c) lim 0 = 0. h h→0+ Maxima and Minima of Functions on Closed Intervals SECTION 3.5 149 Similarly, in the case h < 0, we find that f (c + h) − f (c) 0. h y y = x3 Therefore, f (c) = lim h→0− x f (c + h) − f (c) lim 0 = 0. h h→0− Because both f (c) 0 and f (c) 0, we conclude that f (c) = 0. This establishes ◆ Theorem 2. BEWARE The converse of Theorem 2 is false. That is, the fact that f (c) = 0 is not enough to imply that f (c) is a local extremum. For example, consider the function f (x) = x 3 . Its derivative f (x) = 3x 2 is zero at x = 0. But a glance at its graph (Fig. 3.5.7) shows us that f (0) is not a local extremum of f . FIGURE 3.5.7 There is no extremum at x = 0 even though the derivative is zero there. Thus the equation f (c) = 0 is a necessary condition for f (c) to be a local maximum or minimum value for a function f that is differentiable on an open interval containing c. It is not a sufficient condition. The reason: f (x) may well be zero at points other than local maxima and minima. We give sufficient conditions for local maxima and minima in Chapter 4. The Closed-Interval Maximum-Minimum Method y Global maximum Local, not global x Local, not global Global minimum In most types of optimization problems, we are less interested in the local extrema (as such) than in the absolute, or global, maximum and minimum values attained by a given continuous function. If f is a function with domain D, we call f (c) the absolute maximum value, or global maximum value, of f on D provided that f (c) f (x) for all x in D. Briefly, f (c) is the largest value of f on D. It should be clear how the global minimum of f is to be defined. Figure 3.5.8 illustrates some local and global extrema. On the one hand, every global extremum is, of course, local as well. On the other hand, the graph shows local extrema that are not global. Theorem 3 tells us that the absolute maximum and absolute minimum values of the continuous function f on the closed interval [a, b] occur either at one of the endpoints a or b or at a critical point of f . The number c in the domain of f is called a critical point of f if either • • FIGURE 3.5.8 Some extrema are global; others are merely local. f (c) = 0, or f (c) does not exist. THEOREM 3 Absolute Maxima and Minima Suppose that f (c) is the absolute maximum (or absolute minimum) value of the continuous function f on the closed interval [a, b]. Then c is either a critical point of f or one of the endpoints a and b. Proof This result follows almost immediately from Theorem 2. If c is not an endpoint of [a, b], then f (c) is a local extremum of f on the open interval (a, b). In this ◆ case Theorem 2 implies that f (c) = 0, provided that f is differentiable at c. As a consequence of Theorem 3, we can find the (absolute) maximum and minimum values of the function f on the closed interval [a, b] as follows: 1. Locate the critical points of f : those points where f (x) = 0 and those points where f (x) does not exist. 2. List the values of x that yield possible extrema of f : the two endpoints a and b and those critical points that lie in [a, b]. 3. Evaluate f (x) at each point in this list of possible extrema. 4. Inspect these values of f (x) to see which is the smallest and which is the largest. 149 150 CHAPTER 3 The Derivative The largest of the values in Step 4 is the absolute maximum value of f ; the smallest, the absolute minimum. We call this procedure the closed-interval maximum-minimum method. EXAMPLE 3 For our final discussion of the animal pen problem, let us apply the closed-interval maximum-minimum method to find the maximum and minimum values of the differentiable function f (x) = 35 x(30 − x) = 35 (30x − x 2 ) on the closed interval [0, 30]. Solution The derivative of f is f (x) = 35 (30 − 2x), which is zero only at the point x = 15 in [0, 30]. Including the two endpoints, our list of the only values of x that can yield extrema of f consists of 0, 15, and 30. We evaluate f at each: f (0) = 0, ←− absolute minimum f (15) = 135, ←− absolute maximum f (30) = 0. ←− absolute minimum Thus the maximum value of f (x) on [0, 30] is 135 (attained at x = 15), and the minimum value is 0 (attained both at x = 0 and at x = 30). ◗ EXAMPLE 4 Find the maximum and minimum values of f (x) = 2x 3 − 3x 2 − 12x + 15 on the closed interval [0, 3]. Solution The derivative of f is f (x) = 6x 2 − 6x − 12 = 6(x − 2)(x + 1). So the critical points of f are the solutions of the equation 6(x − 2)(x + 1) = 0 and the numbers c for which f (c) does not exist. There are none of the latter, so the critical points of f occur at x = −1 and x = 2. The first of these is not in the domain of f ; we discard it, and thus the only critical point of f in [0, 3] is x = 2. Including the two endpoints, our list of all values of x that yield a possible maximum or minimum value of f consists of 0, 2, and 3. We evaluate the function f at each: f (0) = 15, ←− absolute maximum f (2) = −5, ←− absolute minimum f (3) = 6. Therefore the maximum value of f on [0, 3] is f (0) = 15 and its minimum value is f (2) = −5. ◗ If in Example 4 we had asked for the maximum and minimum values of f (x) on the interval [−2, 3] (instead of the interval [0, 3]), then we would have included both critical points x = −1 and x = 2 in our list of possibilities. The resulting values of f would have been f (−2) = 11, f (−1) = 22, ←− absolute maximum f (2) = −5, ←− absolute minimum f (3) = 6. 150 Maxima and Minima of Functions on Closed Intervals SECTION 3.5 151 Figure 3.5.9 shows both the curve y = f (x) and the graph of its derivative. Note the vertical line segments joining high and low points on y = f (x) with x-intercepts of dy/d x = f (x). Thus the figure illustrates the following fact: 30 (−1, 22) 20 The critical points of a differentiable function f (x) are the zeros of its derivative f (x). y = f(x) 10 y On the basis of this principle, we can approximate a critical point of f graphically by “zooming in” on a zero of f . In Example 4 the function f was differentiable everywhere. Examples 5 and 6 illustrate the case of an extremum at a critical point where the function is not differentiable. 0 (2, −5) −10 y = f '(x) −20 −2 −1 0 1 2 x EXAMPLE 5 Find the maximum and minimum values of the function f (x) = 3 − |x − 2| on the interval [1, 4]. FIGURE 3.5.9 The critical points of the differentiable function f (x) are the zeros of f (x). Solution If x 2, then x − 2 0, so f (x) = 3 − (2 − x) = x + 1. If x 2, then x − 2 0, so f (x) = 3 − (x − 2) = 5 − x. y (2, 3) 3 2 y = f(x) = 3 − |x − 2 |, 1< =x< =4 Consequently, the graph of f looks like the one shown in Fig. 3.5.10. The only critical point of f in [1, 4] is the point x = 2, because f (x) takes on only the two values +1 and −1 (and so is never zero), and f (2) does not exist. (Why not?) Evaluation of f at this critical point and at the two endpoints yields f (1) = 2, (1, 2) 1 (4, 1) 1 2 3 4 f (2) = 3, ←− absolute maximum f (4) = 1. ←− absolute minimum ◗ x EXAMPLE 6 Find the maximum and minimum values of FIGURE 3.5.10 Graph of the function of Example 5. f (x) = 5x 2/3 − x 5/3 on the closed interval [−1, 4]. Solution Differentiating f yields f (x) = 10 −1/3 5 2/3 5 −1/3 5(2 − x) − x = x (2 − x) = . x 3 3 3 3x 1/3 Hence f has two critical points in the interval: x = 2, where f (x) = 0, and x = 0, where f (x) does not exist (the graph of f has a vertical tangent at (0, 0)). When we evaluate f at these two critical points and at the two endpoints, we get y (−1, 6) ←− absolute maximum f (0) = 0 ←− absolute minimum f (2) = 5 · 22/3 − 25/3 ≈ 4.76, 6 (2, 4.76) f(x) = 5x2/3 − x5/3 4 2 −2 f (−1) = 6, (4, 2.52) (0, 0) 2 4 x FIGURE 3.5.11 Graph of the function of Example 6. f (4) = 5 · 42/3 − 45/3 ≈ 2.52. Thus the maximum value f (−1) = 6 occurs at an endpoint. The minimum value ◗ f (0) = 0 occurs at a point where f is not differentiable. By using a graphics calculator or computer with graphics capabilities, you can verify that the graph of the function f of Example 6 is that shown in Fig. 3.5.11. But in the usual case of a continuous function that has only finitely many critical points in a given closed interval, the closed-interval maximum-minimum method suffices to determine its maximum and minimum values without requiring any detailed knowledge of the graph of the function. 151 152 CHAPTER 3 The Derivative EXAMPLE 7 Figure 3.5.12 shows the graphs of the function f (x) = 4x 4 − 11x 2 − 5x − 3 30 and its derivative y = f(x) f (x) = 16x 3 − 22x − 5 20 10 y 0 −10 −20 −30 −3 y = f'(x) −2 −1 0 x 1 FIGURE 3.5.12 The graphs y = f (x) and y = f (x). 2 3 in the viewing window −3 x 3, −30 y 30. Evidently the maximum value of f (x) on the closed interval [−2, 2] is the left-endpoint value f (−2) = 27. The lowest point on the graph of y = f (x) and the corresponding zero of its derivative dy/d x = f (x) lie within the small boxes in the figure. To find this lowest point exactly we would need to solve the cubic equation 16x 3 − 22x − 5 = 0. But the lowest point also can be located approximately by using a graphing calculator or computer to zoom in more closely. If we attempt to zoom in on the lowest point without changing the “range factors” or “aspect ratios” of the viewing window, we get a picture like the one in Fig. 3.5.13. Here the magnified graph is indistinguishable from its horizontal tangent line at the low point, so it’s impossible to gauge accurately the x-coordinate of the critical point. Consequently, it is much more effective to zoom in on the corresponding zero of the derivative f (x). We can then locate the indicated critical point with much greater precision. Thus it is clear in Fig. 3.5.14 that the minimum value attained by f (x) on ◗ [−2, 2] is approximately f (1.273) ≈ −16.686. 0.04 −16.682 −16.684 y −16.686 0.02 y = f (x) y = f '(x) y 0 −16.688 −0.02 −16.69 1.272 1.273 1.274 x FIGURE 3.5.13 Zooming in on the minimum shown in Fig. 3.5.12. 1.27 1.272 x 1.274 FIGURE 3.5.14 Zooming in instead on the zero of f (x) shown in Fig. 3.5.12. 3.5 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If f (c) f (x) for all x in the interval [a, b], then f (c) is the minimum value of f on [a, b]. 2. If f is continuous on [a, b], then f has a maximum value on [a, b]. 3. If f (c) f (x) for all x both in the domain of f and in some open interval I , then f (x) is said to be a local maximum value of f . 4. Every local extremum of the function f occurs at a point where f (x) = 0. 5. If f (c) is a local extremum of the function f , then either f (c) = 0 or f (c) does not exist. 6. If f (c) is a local extremum of the function f and c is not an endpoint of the domain of f , then either f (c) = 0 or f (c) does not exist. 7. If f (c) f (x) for every number x in the domain D of the function f , then f (c) is called the global maximum value (or the absolute maximum value) of f on D. 8. The absolute maximum value of f (x) = 2x 3 − 3x 2 − 12x + 15 on [0, 3] is f (0) = 15. 9. The absolute maximum value of f (x) = 3 − |x − 2| on the interval [1, 4] is f (4) = 1. 152 Maxima and Minima of Functions on Closed Intervals SECTION 3.5 153 10. If f ( p) and f (q) are both absolute minimum values of f on its domain, then f ( p) = f (q). 3.5 CONCEPTS: QUESTIONS AND DISCUSSION 1. Suppose that the function f is continuous on the closed interval [a, b]. In each of the five following cases, sketch a possible graph (if any) of f . (a) f has a single critical point c but neither a local minimum nor a local maximum in the open interval (a, b). Discuss the possibility that f is not differentiable at c and the possibility that f is differentiable there. (b) f has two critical points but only a single local extremum in (a, b). (c) f has both a local maximum and a local minimum, but only one critical point in (a, b). (d) f has exactly one local maximum, exactly one local minimum, and exactly three critical points in (a, b). (e) f has three local maxima but only a single local minimum in (a, b). 2. Can you give an example of a polynomial of odd degree that has neither a local minimum value nor a local maximum value? Can you give an example of a polynomial of even degree that has neither an absolute minimum value nor an absolute maximum value? 3. Assume that you have located a point on the graph of a differentiable function where a local extremum occurs. Suppose that you zoom in on this point with a graphing calculator or computer, magnifying at each step by the same factor in the x-direction and the y-direction. Should the graph always look like a horizontal line (as in Fig. 3.5.13) after zooming in sufficiently closely in this manner? 3.5 PROBLEMS In Problems 1 through 10, state whether the given function attains a maximum value or a minimum value (or both) on the given interval. [Suggestion: Begin by sketching a graph of the function.] 1. f (x) = 1 − x; 2. f (x) = 2x + 1; [−1, 1) [−1, 1) 3. f (x) = |x| ; (−1, 1) 1 4. f (x) = √ ; (0, 1] x 5. f (x) = |x − 2| ; (1, 4] 6. f (x) = 5 − x 2 ; [−1, 2) 7. f (x) = x 3 + 1; [−1, 1] 1 ; (−∞, ∞) 8. f (x) = 2 x +1 1 9. f (x) = ; [2, 3] x(1 − x) 1 10. f (x) = ; (0, 1) x(1 − x) In Problems 11 through 40, find the maximum and minimum values attained by the given function on the indicated closed interval. 11. f (x) = 3x − 2; [−2, 3] 12. f (x) = 4 − 3x; [−1, 5] 13. h(x) = 4 − x 2 ; [1, 3] 14. f (x) = x 2 + 3; [0, 5] 15. g(x) = (x − 1) ; [−1, 4] 2 16. h(x) = x + 4x + 7; [−3, 0] 2 17. f (x) = x − 3x; [−2, 4] 3 18. g(x) = 2x − 9x + 12x; 3 [0, 4] 2 4 19. h(x) = x + ; [1, 4] x 16 20. f (x) = x 2 + ; [1, 3] x 21. f (x) = 3 − 2x; [−1, 1] 22. f (x) = x 2 − 4x + 3; [0, 2] 23. f (x) = 5 − 12x − 9x 2 ; 24. f (x) = 2x 2 − 4x + 7; [−1, 1] [0, 2] 25. f (x) = x 3 − 3x 2 − 9x + 5; 26. f (x) = x 3 + x; [−1, 2] 27. f (x) = 3x 5 − 5x 3 ; 28. f (x) = |2x − 3| ; [−2, 4] [−2, 2] [1, 2] 29. f (x) = 5 + |7 − 3x| ; [1, 5] 30. f (x) = |x + 1| + |x − 1| ; [−2, 2] 31. f (x) = 50x − 105x + 72x; 3 1 32. f (x) = 2x + ; 2x 2 [0, 1] [1, 4] 153 154 CHAPTER 3 The Derivative 33. f (x) = x ; x +1 34. f (x) = x ; x2 + 1 35. 36. 37. 38. [0, 3] 49. Fig. 3.5.18 [0, 3] 1−x ; [−2, 5] f (x) = 2 x +3 √ f (x) = 2 − 3 x; [−1, 8] √ f (x) = x 1 − x 2 ; [−1, 1] √ f (x) = x 4 − x 2 ; [0, 2] 39. f (x) = x(2 − x)1/3 ; [1, 3] 40. f (x) = x 1/2 − x 3/2 ; [0, 4] 43. Explain why every real number is a critical point of the greatest integer function f (x) = [[x]]. 44. Prove that every quadratic function f (x) = ax 2 + bx + c (a = 0) has exactly one critical point on the real line. f (x) = ax 3 + bx 2 + cx + d y 0 y 0 −2 −2 −4 2 2 y 0 y 0 −2 −2 −4 −4 4 51. Fig. 3.5.20 52. Fig. 3.5.21 4 4 2 2 y 0 y 0 −2 −2 −4 0 x 4 FIGURE 3.5.17 −4 0 x 4 4 2 2 y 0 y 0 −2 −2 −4 0 x −4 4 0 x 4 (b) 4 4 2 2 y 0 y 0 −2 −2 −4 −4 −4 0 x −4 4 (c) 4 4 FIGURE 3.5.21 FIGURE 3.5.20 0 x 4 (d) 4 4 2 2 y 0 y 0 −2 −2 −4 −4 0 x 4 −4 −4 −4 −4 0 x FIGURE 3.5.19 48. Figure 3.5.17 4 −4 4 (a) In Problems 47 through 52, match the given graph of the function with the graph of its derivative f from those in Fig. 3.5.15, parts (a) through (f). 4 0 x FIGURE 3.5.18 (a = 0) 46. Define f (x) to be the distance from x to the nearest integer. What are the critical points of f ? 154 2 −4 can have either two, one, or no critical points on the real line. Produce examples that illustrate each of the three cases. FIGURE 3.5.16 2 −4 45. Explain why the cubic polynomial function 0 x 4 −4 42. Suppose that f is continuous on [a, b] and differentiable on (a, b) and that f (x) is never zero at any point of (a, b). Explain why the maximum and minimum values of f must occur at the endpoints of the interval [a, b]. −4 4 −4 41. Suppose that f (x) = Ax + B is a linear function and that A = 0. Explain why the maximum and minimum values of f on a closed interval [a, b] must occur at the endpoints of the interval. 47. Figure 3.5.16 50. Fig. 3.5.19 0 x (e) FIGURE 3.5.15 4 −4 0 x (f) 4 Maxima and Minima of Functions on Closed Intervals SECTION 3.5 155 In Problems 53 through 60, find good approximations to the maximum and minimum values of the given function on the indicated closed interval by zooming in on the zeros of the derivative. 53. f (x) = x 3 + 3x 2 − 7x + 10; [−2, 2] 54. f (x) = x 3 + 3x 2 − 7x + 10; [−4, 2] 55. f (x) = x 4 − 3x 3 + 7x − 5; [−3, 3] 56. f (x) = x 4 − 5x 3 + 17x − 5; [−3, 3] 57. f (x) = x 4 − 5x 3 + 17x − 5; [0, 2] 58. f (x) = x 5 − 5x 4 − 15x 3 + 17x 2 + 23x; [−1, 1] 59. f (x) = x 5 − 5x 4 − 15x 3 + 17x 2 + 23x; [−3, 3] 60. f (x) = x − 5x − 15x + 17x + 23x; [0, 10] 5 4 3 2 3.5 INVESTIGATION: When Is Your Coffee Cup Stablest? Your car has no cupholder, so you must place your filled coffee cup on the passenger seat beside you when you start out in the morning. Bitter experience has taught you that the cup is least stable—and most prone to spill—when it’s completely full, but becomes more stable as you drink the coffee and thereby lower its level in the cup. Now you’re ready to apply calculus to analyze this phenomenon. Figure 3.5.22 shows a coffee cup partially filled with coffee. We will assume that it is stablest when the centroid of the cup-plus-coffee is lowest. The centroid of a solid cylinder or cylindrical shell is its geometric central point, and the y-coordinate y of the centroid of a composite body consisting of several pieces with masses m 1 , m 2 , and m 3 having centroids with respective y-coordinates y1 , y2 , and y3 is given by T y= Coffee surface H y R FIGURE 3.5.22 Coffee cup partially filled with coffee to depth y. (1) This formula means that y is an average of the y-coordinates y1 , y2 , and y3 of the individual centroids, each weighted by the corresponding mass. The simplified model of the coffee cup shown in Fig. 3.5.22 consists of the following: • B m 1 y1 + m 2 y2 + m 3 y3 . m1 + m2 + m3 • A side surface that is a cylindrical shell with height H , inner radius R, and thickness T , and A bottom that is a solid cylinder with radius R + T and height B. The cup is partially filled with coffee with depth y and density 1 g/cm3 . For instance, let us take H = 8, R = 3, T = 0.5, and B = 1 (all units are centimeters). Assuming also that the density of the material of the cup itself is δ = 1 g/cm3 , we apply Eq. (1) to derive the function f (y) = 87 + 4y 2 , 34 + 8y 0y8 (2) giving the y-coordinate y = f (y) of the centroid of the cup-plus-coffee as a function of the depth y of the coffee in the cup. Figure 3.5.23 shows the graph of the function f . It appears that the centroid is lowest when y = 2, and thus when the cup is about one-quarter filled with coffee. To find when f (y) = 0, you can differentiate the function in (2) and simplify to obtain y 3.5 3 2.5 2 f (y) = 1.5 1 2(4y 2 + 34y − 87) . (4y + 17)2 (3) 0.5 0 2 4 6 y FIGURE 3.5.23 Centroid height f (y) as a function of coffee depth y. Thus you need only solve √ a quadratic equation to see where the numerator is zero: when y = 14 (−17 ± 7 13). The positive solution gives the optimal depth y ≈ 2.0597 cm of the coffee in your cup—just a bit more than a quarter of the height H = 8 cm of the cup. Carry out this analysis with your own favorite coffee cup. Measure its physical dimensions H , R, T , and B. How can you determine the approximate density δ of its material? 155 156 CHAPTER 3 The Derivative 3.6 APPLIED OPTIMIZATION PROBLEMS This section is devoted to applied maximum-minimum problems (like the animal pen problem of Section 1.1) for which the closed-interval maximum-minimum method of Section 3.5 can be used. When we confront such a problem, there is an important first step: We must determine the quantity to be maximized or minimized. This quantity will be the dependent variable in our analysis of the problem. This dependent variable must then be expressed as a function of an independent variable, one that “controls” the values of the dependent variable. If the domain of values of the independent variable—those that are pertinent to the applied problem— is a closed interval, then we may proceed with the closed-interval maximum-minimum method. This plan of attack can be summarized in the following steps: 1. Find the quantity to be maximized or minimized. This quantity, which you should describe with a word or short phrase and label with a descriptive letter, will be the dependent variable. Because it is a dependent variable, it depends on something else; that quantity will be the independent variable. Here we call the independent variable x. 2. Express the dependent variable as a function of the independent variable. Use the information in the problem to write the dependent variable as a function of x. Always draw a figure and label the variables; this is generally the best way to find the relationship between the dependent and independent variables. Use auxiliary variables if they help, but not too many, for you must eventually eliminate them. You must express the dependent variable as a function of the single independent variable x and various constants before you can compute any derivatives. Find the domain of this function as well as its formula. Force the domain to be a closed and bounded interval if possible—if the natural domain is an open interval, adjoin the endpoints if you can. 3. Apply calculus to find the critical points. Compute the derivative f of the function f that you found in Step 2. Use the derivative to find the critical points— where f (x) = 0 and where f (x) does not exist. If f is differentiable everywhere, then its only critical points occur where f (x) = 0. 4. Identify the extrema. Evaluate f at each critical point in its domain and at the two endpoints. The values you obtain will tell you which is the absolute maximum and which is the absolute minimum. Of course, either or both of these may occur at more than one point. 5. Answer the question posed in the original problem. In other words, interpret your results. The answer to the original problem may be something other than merely the largest (or smallest) value of f . Give a precise answer to the specific question originally asked. Observe how we follow this five-step process in Example 1. EXAMPLE 1 A farmer has 200 yd of fence with which to construct three sides of a rectangular pen; an existing long, straight wall will form the fourth side. What dimensions will maximize the area of the pen? y x Area A = xy x Solution We want to maximize the area A of the pen shown in Fig. 3.6.1. To get a formula for the dependent variable A, we observe that the area of a rectangle is the product of its base and its height. So we let x denote the length of each of the two sides of the pen perpendicular to the wall. We also let y denote the length of the side parallel to the wall. Then the area of the rectangle is given by the formula Wall FIGURE 3.6.1 The rectangular pen of Example 1. A = x y. Now we need to write A as a function of either x or y. Because all 200 yd of fence are to be used, 2x + y = 200, 156 so y = 200 − 2x (1) Applied Optimization Problems SECTION 3.6 157 (We chose to express y in terms of x merely because the algebra is slightly simpler.) Next, we substitute this value of y into the formula A = x y to obtain A(x) = x(200 − 2x) = 200x − 2x 2 . x x (2) This equation expresses the dependent variable A as a function of the independent variable x. Before proceeding, we must find the domain of the function A. It is clear from Fig. 3.6.2 that 0 < x < 100. But to apply the closed-interval maximum-minimum method, we need a closed interval. In this example, we may adjoin the endpoints to (0, 100) to get the closed interval [0, 100]. The values x = 0 and x = 100 correspond to “degenerate” pens of area zero. Because zero cannot be the maximum value of A, there is no harm in thus enlarging the domain of the function A. Now we compute the derivative of the function A in Eq. (2): y 200 FIGURE 3.6.2 The relation in Eq. (1) between x and y (Example 1). dA = 200 − 4x. dx Because A is differentiable, its only critical points occur when dA = 0; dx that is, when 200 − 4x = 0. So x = 50 is the only critical point in the interval (0, 100). Including the endpoints, the extrema of A can occur only at x = 0, 50, and 100. We evaluate A at each: 100 yd 5000 yd2 50 yd A(0) = 0, 50 yd A(50) = 5000, ←− absolute maximum A(100) = 0. Wall FIGURE 3.6.3 The pen with maximal area of Example 1. EXAMPLE 2 A piece of sheet metal is rectangular, 5 ft wide and 8 ft long. Congruent squares are to be cut from its four corners. The resulting piece of metal is to be folded and welded to form an open-topped box (Fig. 3.6.4). How should this be done to get a box of largest possible volume? Cut lines x Fold lines 5 – 2x x 8 – 2x x x x 5 – 2x 8 – 2x FIGURE 3.6.4 Making the box of Example 2. x Thus the maximal area is A(50) = 5000 (yd2 ). From Eq. (1) we find that y = 100 when x = 50. Therefore, for the pen to have maximal area, each of the two sides perpendicular to the wall should be 50 yd long and the side parallel to the wall should be 100 yd long (Fig. 3.6.3). ◗ x 5 FIGURE 3.6.5 The 5-ft width of the metal sheet (Example 2). Solution The quantity to be maximized—the dependent variable—is the volume V of the box to be constructed. The shape and thus the volume of the box are determined by the length x of the edge of each corner square removed. Hence x is a natural choice for the independent variable. To write the volume V as a function of x, note that the finished box will have height x and its base will measure 8 − 2x ft by 5 − 2x ft. Hence its volume is given by V (x) = x(5 − 2x)(8 − 2x) = 4x 3 − 26x 2 + 40x. The procedure described in this example will produce an actual box only if 0 < x < 2.5 (Fig. 3.6.5). But we make the domain the closed interval [0, 2.5] to ensure that a maximum of V (x) exists and to use the closed-interval maximum-minimum method. The values x = 0 and x = 2.5 correspond to “degenerate” boxes of zero volume, so adjoining these points to (0, 2.5) will affect neither the location of the absolute maximum nor its value. Now we compute the derivative of V : V (x) = 12x 2 − 52x + 40 = 4(3x − 10)(x − 1). The only critical points of the differentiable function V occur where 157 158 CHAPTER 3 The Derivative V (x) = 0; that is, where 4(3x − 10)(x − 1) = 0. . We discard the latter because it The solutions of this equation are x = 1 and x = 10 3 does not lie in the domain [0, 2.5] of V . So we examine these values of V : V (0) = 0, V (1) = 18, 1 ft 6 ft V (2.5) = 0. 3 ft FIGURE 3.6.6 The box with maximal volume of Example 2. ←− absolute maximum Thus the maximum value of V (x) on [0, 2.5] is V (1) = 18. The answer to the question posed is this: The squares cut from the corners should be of edge length 1 ft each. The resulting box will measure 6 ft by 3 ft by 1 ft, and its volume will be 18 ft3 (Fig. 3.6.6). ◗ For our next application of the closed-interval maximum-minimum method, let us consider a typical problem in business management. Suppose that x units of computer diskettes are to be manufactured at a total cost of C(x) dollars. We make the simple (but not always valid) assumption that the cost function C(x) is the sum of two terms: • • A constant term a representing the fixed cost of acquiring and maintaining production facilities (overhead), and A variable term representing the additional cost of making x units at, for example, b dollars each. Then the total cost is the sum of the fixed cost and the additional cost, so the cost function C(x) is given by C(x) = a + bx. (3) We also assume that the number of units that can be sold (and hence will be manufactured) is a linear function of the selling price p, so that x = m − np where m and n are positive constants. The minus sign indicates that an increase in selling price will result in a decrease in sales. If we solve this last equation for p, we get the price function p(x) = A − Bx (4) (A and B are also constants). The quantity to be maximized is profit, given here by the profit function P(x), which is equal to the sales revenue minus the production costs. Thus P(x) = x p(x) − C(x). (5) EXAMPLE 3 Suppose that the cost of publishing a small book is $10,000 to set up the (annual) press run plus $8 for each book printed. The publisher sold 7000 copies last year at $13 each, but sales dropped to 5000 copies this year when the price was raised to $15 per copy. Assume that up to 10,000 copies can be printed in a single press run. How many copies should be printed, and what should be the selling price of each copy, to maximize the year’s profit on this book? Solution The dependent variable to be maximized is the profit P. As independent variable we choose the number x of copies to be printed; also, 0 x 10,000. The given cost information then implies that C(x) = 10,000 + 8x. 158 Applied Optimization Problems SECTION 3.6 159 Now we substitute into Eq. (4) the data x = 7000 when p = 13 as well as the data x = 5000 when p = 15. We obtain the equations A − 7000B = 13, A − 5000B = 15. When we solve these equations simultaneously, we find that A = 20 and B = 0.001. Hence the price function is x , p(x) = 20 − 1000 and thus the profit function is x P(x) = x 20 − − (10,000 + 8x). 1000 We expand and collect terms to obtain P(x) = 12x − x2 − 10,000, 1000 0 x 10,000. Now dP x = 12 − , dx 500 and the only critical points of the differentiable function P occur when dP = 0; dx that is, when x = 0; x = 12 · 500 = 6000. 500 We check P at this value of x as well as the values of P(x) at the endpoints to find the maximum profit: 12 − P(0) = −10,000, P(6000) = 26,000, ←− absolute maximum P(10,000) = 10,000. Therefore, the maximum possible annual profit of $26,000 results from printing 6000 copies of the book. Each copy should be sold for $14, because p(6000) = 20 − ◗ EXAMPLE 4 We need to design a cylindrical can with radius r and height h. The top and bottom must be made of copper, which will cost 2/ c/in.2 The curved side is to be 2 made of aluminum, which will cost 1/ c/in. We seek the dimensions that will maximize the volume of the can. The only constraint is that the total cost of the can is to be 300π c/. 2¢/in.2 1¢/in.2 6000 = 14. 1000 h Solution We need to maximize the volume V of the can, which we can compute if we know its radius r and its height h (Fig. 3.6.7). With these dimensions, we find that V = πr 2 h, r FIGURE 3.6.7 The cylindrical can of Example 4. (6) but we need to express V as a function of r alone (or as a function of h alone). Both the circular top and bottom of the can have area πr 2 in.2 , so the area of copper to be used is 2πr 2 and its cost is 4πr 2 cents. The area of the curved side of the can is 2πr h in.2 , so the area of aluminum used is the same, and the aluminum costs 2πr h cents. We obtain the total cost of the can by adding the cost of the copper to the cost of the aluminum. This sum must be 300π c/, and therefore 4πr 2 + 2πr h = 300π. (7) 159 160 CHAPTER 3 The Derivative We eliminate h in Eq. (6) by solving Eq. (7) for h: h= 1 300π − 4πr 2 = (150 − 2r 2 ). 2πr r (8) Hence 1 V = V (r ) = (πr 2 ) (150 − 2r 2 ) = 2π(75r − r 3 ). r (9) To determine the √ domain of definition of V , we note Eq. (7) that √ from √ 4πr < 300π , so r < 75 for the desired can; with r = 75 = 5 3, we get a degenerate can with height h = 0. With r = 0, we obtain no value of h in Eq. (8) and therefore no can, but V (r ) is√nevertheless continuous at r = 0. Consequently, we can take the closed interval [0, 5 3] to be the domain of V . Calculating the derivative yields 2 V (r ) = 2π(75 − 3r 2 ) = 6π(25 − r 2 ). Because V (r ) is a polynomial, V (r ) exists for all values of r , so we obtain all critical points by solving the equation V (r ) = 0; that is, 6π(25 − r 2 ) = 0. 20 We discard the solution −5, as it does√not lie in the domain of V . Thus we obtain only the single critical point r = 5 in [0, 5 3]. Now V (0) = 0, V (5) = 500π, √ V 5 3 = 0. 5 FIGURE 3.6.8 The can of maximal volume in Example 4. ←− absolute maximum Thus the can of maximum volume has radius r = 5 in., and Eq. (8) yields its height to ◗ be h = 20 in. Figure 3.6.8 shows such a can. EXAMPLE 5 (A Sawmill Problem) Suppose that you need to cut a beam with maximal rectangular cross section from a circular log of radius 1 ft. (This is the geometric problem of finding the rectangle of greatest area that can be inscribed in a circle of radius 1.) What are the shape and cross-sectional area of such a beam? Radius 1 2y x 2x y Solution Let x and y denote half the base and half the height, respectively, of the inscribed rectangle (Fig. 3.6.9). Apply the Pythagorean theorem to the small right triangle in the figure. This yields the equation so y = 1 − x 2 . x 2 + y 2 = 1, The area of the inscribed rectangle is A = (2x)(2y) = 4x y. You may now express A as a function of x alone: A(x) = 4x 1 − x 2 . The practical domain of definition of A is (0, 1), and there is no harm (and much advantage) in adjoining the endpoints, so you take [0, 1] to be the domain. Next, FIGURE 3.6.9 A sawmill problem—Example 5. dA 4 − 8x 2 = 4 · (1 − x 2 )1/2 + 2x(1 − x 2 )−1/2 (−2x) = . dx (1 − x 2 )1/2 You observe that A (1) does not exist, but this causes no trouble, because differentiability at the endpoints is not assumed in Theorem 3 of Section 3.5. Hence you need only solve the equation A (x) = 0; 160 Applied Optimization Problems SECTION 3.6 161 that is, 4 − 8x 2 = 0. √ 1 − x2 A fraction can be zero only when its numerator is zero and its denominator is not, so A (x) = 0 when 4 − 8x 2 = 0. Thus√you find the only critical √ point of A in the open √ 1 interval (0, 1) to be x = 1/2 = 2 2 (and 2x = 2y = 2). You evaluate A here and at the two endpoints to find that A(0) = 0, √ A 12 2 = 2, ←− absolute maximum A(1) = 0. Therefore, √ the beam with rectangular cross section of maximal area is square, with ◗ edges 2 ft long and with cross-sectional area 2 ft2 . 2 2 FIGURE 3.6.10 Cut four more beams after cutting one large beam. In Problem 43 we ask you to maximize the total cross-sectional area of the four planks that can be cut from the four pieces of log that remain after cutting the square beam (Fig. 3.6.10). Plausibility You should always check your answers for plausibility. In Example 5, the cross-sectional area of the log from which the beam is to be cut is π ≈ 3.14 ft2 . The beam of maximal cross-section area 2 ft2 thus uses a little less than 64% of the log. This is plausible. Had the fraction been an extremely inefficient 3% or a wildly optimistic 98%, you should have searched for an error in arithmetic, algebra, calculus, or logic (as you would had the fraction been −14% or 150%). Check the results of Examples 1 through 4 for plausibility. Dimensions Another way to check answers is to use dimensional analysis. Work the problem with unspecified constants in place of the actual numbers. In Example 5, it would be good practice to find the beam of maximal rectangular cross section that can be cut from a circular log of radius R rather than radius 1 ft. You can always substitute the given value R = 1 at the conclusion of the solution. A brief solution to this problem might go as follows: Dimensions of beam: base 2x, height 2y. Area of beam: A = 4x y. Draw a radius of the log from its center to one corner of the rectangular beam, as in Fig. 3.6.11. This radius has length R, so the Pythagorean theorem gives y = R2 − x 2. x 2 + y2 = R2; R y x FIGURE 3.6.11 The log with radius R. 161 162 CHAPTER 3 The Derivative Area of beam: A(x) = 4x R 2 − x 2 , 0 x R. 4R 2 − 8x 2 A (x) = 4(R 2 − x 2 )1/2 + 2x(R 2 − x 2 )−1/2 (−2x) = √ . R2 − x 2 Area A = 2R2 R 2 R 2 A(R) = 0. FIGURE 3.6.12 The inscribed square beam with maximal cross-sectional area. Figure 3.6.12 shows the dimensions of the inscribed rectangle of maximal area. Now you can check the results for dimensional accuracy. The value of x that maximizes A is a length (R) multiplied by a pure (dimensionless) numerical constant √ ( 12 2), so x has the dimensions of length—that’s correct; had it been anything else, you would need to search for the error. Moreover, the maximum cross-sectional area of the beam is 2R 2 , the product of a pure number and the square of a length, thus having the dimensions of area. This, too, is correct. A B d1 d2 a b α β A' M c−x x P c A (x) does not exist when x = R, but that’s an endpoint; we’ll check it separately. √ A (x) = 0 when x = 12 R 2 (ignore the negative root; it’s not in the domain of A). A(0) = 0, √ ←− absolute maximum A 12 R 2 = 2R 2 , B' FIGURE 3.6.13 Reflection at P of a light ray by a mirror M (Example 6). EXAMPLE 6 We consider the reflection of a ray of light by a mirror M as in Fig. 3.6.13, which shows a ray traveling from point A to point B via reflection off M at the point P. We assume that the location of the point of reflection is such that the total distance d1 + d2 traveled by the light ray will be minimized. This is an application of Fermat’s principle of least time for the propagation of light. The problem is to find P. Solution Drop perpendiculars from A and B to the plane of the mirror M. Denote the feet of these perpendiculars by A and B (Fig. 3.6.13). Let a, b, c, and x denote the lengths of the segments A A , B B , A B , and A P, respectively. Then c − x is the length of the segment P B . By the Pythagorean theorem, the distance to be minimized is then (10) d1 + d2 = f (x) = a 2 + x 2 + b2 + (c − x)2 . We may choose as the domain of f the interval [0, c], because the minimum of f must occur somewhere within that interval. (To see why, examine the picture you get if x is not in that interval.) Then x (c − x)(−1) + . f (x) = √ a2 + x 2 b2 + (c − x)2 (11) Recognizing the distances d1 and d2 in the denominators here, we see that f (x) = x c−x − . d1 d2 (12) Consequently, any horizontal tangent to the graph of f must occur over the point x determined by the equation c−x x = . d1 d2 (13) At such a point, cos α = cos β, where α is the angle of the incident light ray and β is the angle of the reflected ray (Fig. 3.6.13). Both α and β lie between 0 and π/2, and thus we find that α = β. In short, the point P must be located so that the angle of ◗ incidence is equal to the angle of reflection, a familiar principle from physics. 162 Applied Optimization Problems SECTION 3.6 163 The computation in Example 6 has an alternative interpretation that is interesting, if somewhat whimsical. Figure 3.6.14 shows a feedlot 200 ft long with a water trough along one edge and a feed bin located on an adjacent edge. A cow enters the gate at the point A, 90 ft from the water trough. She walks straight to point P, gets a drink from the trough, and then walks straight to the feed bin at point B, 60 ft from the trough. If the cow knew calculus, what point P along the water trough would she select to minimize the total distance she walks? A Feed bin B 90 60 P Water trough x 200 – x FIGURE 3.6.14 The feedlot. In comparing Figs. 3.6.13 and 3.6.14, we see that the cow’s problem is to minimize the distance function f in Eq. (10) with the numerical values a = 90, b = 60, and c = 200. When we substitute these values and d1 = a 2 + x 2 and d2 = b2 + (c − x)2 in Eq. (13), we get √ x 8100 + x2 = 200 − x 3600 + (200 − x)2 . We square both sides, clear the equation of fractions, and simplify. The result is x 2 [3600 + (200 − x)2 ] = (200 − x)2 (8100 + x 2 ); 3600x 2 60x 150x x = 8100(200 − x)2 ; = 90(200 − x); = 18,000; = 120. (Why?) Thus the cow should proceed directly to the point P located 120 ft along the water trough. These examples indicate that the closed-interval maximum-minimum method is applicable to a wide range of problems. Indeed, applied optimization problems that seem as different as light rays and cows may have essentially identical mathematical models. This is only one illustration of the power of generality that calculus exploits so effectively. 3.6 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The maximum area of the pen of Example 1 is 5000 square yards. 2. The maximum area of the pen of Example 1 occurs when the side parallel to the wall has length 100 yards. 3. The domain of the volume function of Example 2 is determined by the fact that neither the length, nor the width, nor the height of the box can be negative. √ 4. In Example 4 we went to some trouble to obtain the closed interval [0, 5 3] for the domain of the function V because we had nothing better to do. 5. In Example 5 the area function A(x) is not differentiable at the endpoint x = 1 of its domain, so the area function has no extremum there. 163 164 CHAPTER 3 The Derivative 6. It is plausible that when the rectangle of largest possible area is inscribed in a circle, then the rectangle occupies 3% of the area of the circle. 7. It is reasonable that when the rectangle of largest possible area is inscribed in a circle of radius R, then the area of the rectangle is 2R ft. 8. Light travels from point A to point B in such a way to minimize the total time to get from A to B. 9. To solve an applied maximum-minimum problem, it is usually wise to begin by identifying the quantity to be maximized or minimized. √ 10. To solve an equation such as 8 − x 2 = x, it is usually wise to begin by squaring both sides in order to eliminate the radical. 3.6 CONCEPTS: QUESTIONS AND DISCUSSION 1. How do you decide what is the dependent variable in an optimization problem? The independent variable? Discuss the differences in the roles played by dependent and independent variables in an optimization problem. 2. Discuss the differences among the following three items: • • • A relation among two or more variables describing an applied problem. A formula giving the dependent variable in terms of other variables. A function expressing the dependent variable in terms of an independent variable. Outline and contrast the roles played by relations, formulas, and functions in typical optimization problems. 3.6 PROBLEMS 1. Find two positive real numbers x and y such that their sum is 50 and their product is as large as possible. 2. Find the maximum possible area of a rectangle of perimeter 200 m. 5. A rectangular box has a square base with edges at least 1 in. long. It has no top, and the total area of its five sides is 300 in.2 (Fig. 3.6.16). What is the maximum possible volume of such a box? 3. A rectangle with sides parallel to the coordinate axes has one vertex at the origin, one on the positive x-axis, one on the positive y-axis, and its fourth vertex in the first quadrant on the line with equation 2x + y = 100 (Fig. 3.6.15). What is the maximum possible area of such a rectangle? y y x x FIGURE 3.6.16 A box with square base and volume V = x 2 y (Problems 5, 17, and 20). (x, y) 2x + y = 100 x 164 6. If x is in the interval [0, 1], then x − x 2 is not negative. What is the maximum value that x − x 2 can have on that interval? In other words, what is the greatest amount by which a real number can exceed its square? FIGURE 3.6.15 The rectangle of Problem 3. 7. The sum of two positive numbers is 48. What is the smallest possible value of the sum of their squares? 4. A farmer has 600 m of fencing with which to enclose a rectangular pen adjacent to a long existing wall. He will use the wall for one side of the pen and the available fencing for the remaining three sides. What is the maximum area that can be enclosed in this way? 8. A rectangle of fixed perimeter 36 is rotated around one of its sides, thus sweeping out a figure in the shape of a right circular cylinder (Fig. 3.6.17). What is the maximum possible volume of that cylinder? Applied Optimization Problems SECTION 3.6 165 16. What is the maximum possible area of a rectangle with a base that lies on the x-axis and with two upper vertices that lie on the graph of the equation y = 4 − x 2 (Fig. 3.6.20)? h y y = 4 − x2 r (x, y) FIGURE 3.6.17 The rectangle and cylinder of Problem 8. 9. The sum of two nonnegative real numbers is 10. Find the minimum possible value of the sum of their cubes. 10. Suppose that the strength of a rectangular beam is proportional to the product of the width and the square of the height of its cross section. What shape beam should be cut from a cylindrical log of radius r to achieve the greatest possible strength? 11. A farmer has 600 yd of fencing with which to build a rectangular corral. Some of the fencing will be used to construct two interval divider fences, both parallel to the same two sides of the corral (Fig. 3.6.18). What is the maximum possible total area of such a corral? y y y y x FIGURE 3.6.18 The divided corral of Problem 11. x FIGURE 3.6.20 The rectangle of Problem 16. 17. A rectangular box has a square base with edges at least 1 cm long. Its total surface area is 600 cm2 . What is the largest possible volume that such a box can have? 18. You must make a cylindrical can with a bottom but no top from 300π in.2 of sheet metal. No sheet metal will be wasted; you are allowed to order a circular piece of any size for its base and any appropriate rectangular piece to make into its curved side so long as the given conditions are met. What is the greatest possible volume of such a can? 19. Three large squares of tin, each with edges 1 m long, have four small, equal squares cut from their corners. All twelve resulting small squares are to be the same size (Fig. 3.6.21). The three large cross-shaped pieces are then folded and welded to make boxes with no tops, and the twelve small squares are used to make two small cubes. How should this be done to maximize the total volume of all five boxes? x x 12. Find the maximum possible volume of a right circular cylinder if its total surface area—including both circular ends—is 150π . 13. Find the maximum possible area of a rectangle with diagonals of length 16. 14. A rectangle has a line of fixed length L reaching from one vertex to the midpoint of one of the far sides (Fig. 3.6.19). What is the maximum possible area of such a rectangle? L y 2 x FIGURE 3.6.19 The rectangle of Problem 14. 15. The volume V (in cubic centimeters) of 1 kg of water at temperature T between 0◦ C and 30◦ C is very closely approximated by V = 999.87 − (0.06426)T + (0.0085043)T 2 − (0.0000679)T 3 . At what temperature does water have its maximum density? 1m x x 1m FIGURE 3.6.21 One of the three 1-m squares of Problem 19. 20. Suppose that you are to make a rectangular box with a square base from two different materials. The material for the top and four sides of the box costs $1/ft2 ; the material for the base costs $2/ft2 . Find the dimensions of the box of greatest possible volume if you are allowed to spend $144 for the material to make it. 21. A piece of wire 80 in. long is cut into at most two pieces. Each piece is bent into the shape of a square. How should this be done to minimize the sum of the area(s) of the square(s)? To maximize it? 22. A wire of length 100 cm is cut into two pieces. One piece is bent into a circle, the other into a square. Where should the cut be made to maximize the sum of the areas of the square and the circle? To minimize that sum? 23. A farmer has 600 m of fencing with which she plans to enclose a rectangular pasture adjacent to a long existing wall. She plans to build one fence parallel to the wall, two to form 165 166 CHAPTER 3 The Derivative the ends of the enclosure, and a fourth (parallel to the ends of the enclosure) to divide it equally. What is the maximum area that can be enclosed? 24. A zookeeper needs to add a rectangular outdoor pen to an animal house with a corner notch, as shown in Fig. 3.6.22. If 85 m of new fence is available, what dimensions of the pen will maximize its area? No fence will be used along the walls of the animal house. decrease. What fare should be charged to get the largest possible revenue? 32. Find the shape of the cylinder of maximal volume that can be inscribed in a sphere of radius R (Fig. 3.6.23). Show √ that the ratio of the height of the cylinder to its radius is 2 and that the ratio of√ the volume of the sphere to that of the maximal cylinder is 3. Animal house h 2 R r 10 m 5m New fence FIGURE 3.6.22 The rectangular pen of Problem 24. FIGURE 3.6.23 The sphere and cylinder of Problem 32. 33. Find the dimensions of the right circular cylinder of greatest volume that can be inscribed in a right circular cone of radius R and height H (Fig. 3.6.24). 25. Suppose that a post office can accept a package for mailing only if the sum of its length and its girth (the circumference of its cross section) is at most 100 in. What is the maximum volume of a rectangular box with square cross section that can be mailed? H 26. Repeat Problem 25, but use a cylindrical package; its cross section is circular. 27. A printing company has eight presses, each of which can print 3600 copies per hour. It costs $5.00 to set up each press for a run and 10 + 6n dollars to run n presses for 1 h. How many presses should be used to print 50,000 copies of a poster most profitably? 28. A farmer wants to hire workers to pick 900 bushels of beans. Each worker can pick 5 bushels per hour and is paid $1.00 per bushel. The farmer must also pay a supervisor $10 per hour while the picking is in progress, and he has additional miscellaneous expenses of $8 per worker. How many workers should he hire to minimize the total cost? What will then be the cost per bushel picked? 29. The heating and cooling costs for a certain uninsulated house are $500/yr, but with x 10 in. of insulation, the costs are 1000/(2 + x) dollars/yr. It costs $150 for each inch (thickness) of insulation installed. How many inches of insulation should be installed to minimize the total (initial plus annual) costs over a 10-yr period? What will then be the annual savings resulting from this optimal insulation? 30. A concessionaire had been selling 5000 burritos each game night at 50/ c each. When she raised the price to 70/ c each, sales dropped to 4000 per night. Assume a linear relationship between price and sales. If she has fixed costs of $1000 per night and each burrito costs her 25/ c, what price will maximize her nightly profit? 31. A commuter train carries 600 passengers each day from a suburb to a city. It costs $1.50 per person to ride the train. Market research reveals that 40 fewer people would ride the train for each 5/ c increase in the fare, 40 more for each 5/ c 166 h r R FIGURE 3.6.24 The cone and cylinder of Problem 33. 34. Figure 3.6.25 shows a circle of radius 1 in which a trapezoid is inscribed. The longer of the two parallel sides of the trapezoid coincides with a diameter of the circle. What is the maximum possible area of such a trapezoid. (Suggestion: A positive quantity is maximized when its square is maximized.) 2x 1 1 FIGURE 3.6.25 The circle and trapezoid of Problem 34. 35. Show that the rectangle of maximal perimeter that can be inscribed in a circle is a square. Applied Optimization Problems SECTION 3.6 167 36. Find the dimensions of the rectangle (with sides parallel to the coordinate axes) of maximal area that can be inscribed in the ellipse with equation x2 y2 + =1 25 9 (Fig. 3.6.26). y (0, 3) (x, y) (5, 0) 45. A small island is 2 km off shore in a large lake. A woman on the island can row her boat 10 km/h and can run at a speed of 20 km/h. If she rows to the closest point of the straight shore, she will land 6 km from a village on the shore. Where should she land to reach the village most quickly by a combination of rowing and running? 46. A factory is located on one bank of a straight river that is 2000 m wide. On the opposite bank but 4500 m downstream is a power station from which the factory draws its electricity. Assume that it costs three times as much per meter to lay an underwater cable as to lay an aboveground cable. What path should a cable connecting the power station to the factory take to minimize the cost of laying the cable? 47. A company has plants that are located (in an appropriate coordinate system) at the points A(0, 1), B(0, −1), and C(3, 0) (Fig. 3.6.29). The company plans to construct a distribution center at the point P(x, 0). What value of x would minimize the sum of the distances from P to A, B, and C? y FIGURE 3.6.26 The ellipse and rectangle of Problem 36. 37. A right √ circular cone of radius r and height h has slant height L = r 2 + h 2 . What is the maximum possible volume of a cone with slant height 10? A P x 38. Two vertical poles 10 ft apart are both 10 ft tall. Find the length of the shortest rope that can reach from the top of one pole to a point on the ground between them and then to the top of the other pole. B 39. The sum of two nonnegative real numbers is 16. Find the maximum possible value and the minimum possible value of the sum of their cube roots. 40. A straight wire 60 cm long is bent into the shape of an L. What is the shortest possible distance between the two ends of the bent wire? 41. What is the shortest possible distance from a point on the parabola y = x 2 to the point (0, 1)? 42. Given: There is exactly one point on the graph of y = √ 3 3x − 4 that is closest to the origin. Find it. (Suggestion: See Fig. 3.6.27, and solve the equation you obtain by inspection.) 2 FIGURE 3.6.29 The locations in Problem 47. 48. Light travels at speed c in air and at a slower speed v in water. (The constant c is approximately 3 × 1010 cm/s; the ratio n = c/v, known as the index of refraction, depends on the color of the light but is approximate 1.33 for water.) Figure 3.6.30 shows the path of a light ray traveling from point A in air to point B in water, with what appears to be a sudden change in direction as the ray moves through the air-water interface. (a) Write the time T required for the ray to travel from A to B in terms of the variable x and the constants a, b, c, s, and v, all of which have been defined or are shown in the figure. (b) Show that the equation T (x) = 0 for minimizing T is equivalent to the condition 1 y c sin α = = n. sin β v 0 −1 1 1 −2 −2 C −1 0 x 1 FIGURE 3.6.27 The curve of Problem 42. This is Snell’s law: The ratio of the sines of the angles of incidence and refraction is equal to the index of refraction. A 2 1 FIGURE 3.6.28 The rectangle and equilateral triangle of Problem 44. 43. Find the dimensions that maximize the cross-sectional area of the four planks that can be cut from the four pieces of the circular log of Example 5—the pieces that remain after a square beam has been cut from the log (Fig. 3.6.10). 44. Find the maximal area of a rectangle inscribed in an equilateral triangle with edges of length 1, as in Fig. 3.6.28. a P α x s−x β Air Q Water b B FIGURE 3.6.30 Snell’s law gives the path of refracted light (Problem 48). 167 168 CHAPTER 3 The Derivative should the lengths of the inner struts be to maximize the area of the kite? 2 2 4 4 FIGURE 3.6.32 The kite frame (Problem 52). Refraction of light at an air-water interface 49. The mathematics of Snell’s law (Problem 48) is applicable to situations other than the refraction of light. Figure 3.6.31 shows an east-west geologic fault that separates two towns at points A and B. Assume that A is a miles north of the fault, that B is b miles south of the fault, and that B is L miles east of A. We want to build a road from A to B. Because of differences in terrain, the cost of construction is C1 (in millions of dollars per mile) north of the fault and C2 south of it. Where should the point P be placed to minimize the total cost of road construction? (a) Using the notation in the figure, show that the cost is minimized when C1 sin θ1 = C2 sin θ2 . (b) Take a = b = C1 = 1, C2 = 2, and L = 4. Show that the equation in part (a) is equivalent to f (x) = 3x 4 − 24x 3 + 51x 2 − 32x + 64 = 0. To approximate the desired solution of this equation, calculate f (0), f (1), f (2), f (3), and f (4). You should find that f (3) > 0 > f (4). Interpolate between f (3) and f (4) to approximate the desired root of this equation. Problems 53 through 55 deal with alternative methods of constructing a tent. 53. Figure 3.6.33 shows a 20-by-20-ft square of canvas tent material. Girl Scout Troop A must cut pieces from its four corners as indicated, so that the four remaining triangular flaps can be turned up to form a tent in the shape of a pyramid with a square base. How should this be done to maximize the volume of the tent? Let A denote the area of the base of the tent and h its height. With x as indicated in the figure, show that the volume V = 13 Ah of the tent is given by √ V (x) = 43 x 2 100 − 20x, 0 x 5. Maximize V by graphing V (x) and V (x) and zooming in on the zero of V (x). A a θ1 10 − x L−x W E x 10 − x h x x P b θ2 B FIGURE 3.6.33 The canvas square—first attempt. FIGURE 3.6.31 Building a road from A to B (Problem 49). 50. The sum of the volumes of two cubes is 2000 in.3 What should their edges x and y be to maximize the sum of their surface areas? To minimize it? 51. The sum of the surface areas of a cube and a sphere is 1000 in.2 What should their dimensions be to minimize the sum of their volumes? To maximize it? 52. Your brother has six pieces of wood with which to make the kite frame shown in Fig. 3.6.32. The four outer pieces with the indicated lengths have already been cut. How long 168 54. Girl Scout Troop B must make a tent in the shape of a pyramid with a square base from a similar 20-by-20-ft square of canvas but in the manner indicated in Fig. 3.6.34. With x as indicated in the figure, show that the volume of the tent is given by √ V (x) = 23 x 2 200 − 20x, 0 x 10. Maximize V graphically as in Problem 53. Derivatives of Trigonometric Functions SECTION 3.7 169 61. Figure 3.6.35 shows a triangle bounded by the nonnegative coordinate axes and the line tangent to the curve y = 1/(1 + x 2 ) at the first-quadrant point (x, y). Is it apparent that the area A(x) of this triangle is very large when x > 0 is very close to zero? But your task is to find the maximum and minimum values of A for 12 x 2. It will be convenient to use a computer algebra system, both to find A(x) and to solve the sixth-degree equation you should encounter. x FIGURE 3.6.34 The canvas square—second attempt. y 55. Solve Problems 53 and 54 analytically to verify √that the maximal volume in Problem 54 is exactly 2 2 times the maximal volume in Problem 53. It pays to think before making a tent! y = Problems 56 and 57 deal with rectangular boxes with square base. Such a box is said to be closed if it has both a (square) bottom and a top (as well as four vertical sides), open if it has a bottom but no top. 56. Show that, among all closed square-based rectangular boxes with a given fixed total surface area, the one with maximal volume is a cube. 57. Show that, among all open square-based rectangular boxes with a given fixed total surface area, the one with maximal volume has height equal to half the length of the edge of its base. Problems 58 through 60 deal with right circular cylinders. Such a “can” is said to be closed if it has both a (circular) bottom and a top (as well as a curved side), open if it has a bottom but no top. 58. Show that, among all closed cylindrical cans with a given fixed total surface area, the one with maximal volume has height equal to the diameter of its base. 59. Show that, among all open cylindrical cans with a given fixed total surface area, the one with maximal volume has height equal to the radius of its base. 60. Suppose that the bottom and curved side surface of a poptop soft drink can have the same thickness. But, in order that the top not be ripped upon opening, it is three times as thick as the bottom. Show that, among all such soft drink cans made from a fixed total amount of material (including the triple-thick top), the one with maximal volume has height approximately twice its diameter. (Perhaps this is why soft drink cans look somewhat taller than soup or vegetable cans.) Suggestion: To simplify the computations, you may assume that the amount of material used to make a can of inner radius r , inner height h, and thickness t (except for the top, of thickness 3t), is πr 2 t + 2πr ht + 3πr 2 t. This will be quite accurate if t is very small in comparison with r and h. 1 1 + x2 (x, y ) x FIGURE 3.6.35 Triangle bounded by coordinate axes and a tangent line 1 to the curve y = . 1 + x2 62. Figure 3.6.36 shows a one-mile-square city park in central Villabuena. A local power company needs to run a power line from the northwest corner A of the park to the southeast corner B. To preserve the beauty of the park, only underground lines may be run through the park itself, but overhead lines are permissible along the boundary of the park. The power company plans to construct an overhead line a distance x along the west edge of the park, then from the southern end of this line continue with a straight power line to point B. If overhead lines cost $40 thousand per mile and underground lines cost $100 thousand per mile, how should the power company construct the line to minimize its total cost? A x B FIGURE 3.6.36 The onemile-square park in central Villabuena. 3.7 DERIVATIVES OF TRIGONOMETRIC FUNCTIONS In this section we begin our study of the calculus of trigonometric functions, focusing first on the sine and cosine functions. The definitions and the elementary properties of trigonometric functions are reviewed in Appendix C. 169 170 CHAPTER 3 The Derivative When we write sin θ (or cos θ), we mean the sine (or cosine) of an angle of θ radians (rad). Recall the fundamental relation between radian measure and degree measure of angles: π radians = 180 degrees. Radians Degrees 0 π/6 π/4 π/3 π/2 2π/3 3π/4 5π/6 π 3π/2 2π 4π 0 30 45 60 90 120 135 150 180 270 360 720 (1) Upon division of both sides of this equation by π and 180, respectively, and abbreviating the units, we get the conversion relations 1 rad = 180 deg π and 1 deg = π rad. 180 Figure 3.7.1 shows radian-degree conversions for some frequently occurring angles. The derivatives of the sine and cosine functions depend on the limits sin θ = 1, θ→0 θ lim 1 − cos θ =0 θ→0 θ lim that we established in Section 2.3. The addition formulas cos(x + y) = cos x cos y − sin x sin y, sin(x + y) = sin x cos y + cos x sin y FIGURE 3.7.1 Some radian- degree conversions. (2) (3) are needed as well. THEOREM 1 Derivatives of Sines and Cosines The functions f (x) = sin x and g(x) = cos x are differentiable for all x, and Dx sin x = cos x, Dx cos x = − sin x. (4) (5) Proof To differentiate f (x) = sin x, we begin with the definition of the derivative, f (x) = lim h→0 f (x + h) − f (x) sin(x + h) − sin x = lim . h→0 h h Next we apply the addition formula for the sine and the limit laws to get (sin x cos h + sin h cos x) − sin x h→0 h f (x) = lim 1 − cos h sin h − (sin x) h→0 h h sin h 1 − cos h − (sin x) lim . = (cos x) lim h→0 h h→0 h = lim (cos x) The limits in Eq. (2) now yield f (x) = (cos x)(1) − (sin x)(0) = cos x, which proves Eq. (4). The proof of Eq. (5) is quite similar. (See Problem 72.) ◆ Examples 1 through 4 illustrate the application of Eqs. (4) and (5) in conjunction with the general differentiation formulas of Sections 3.2, 3.3, and 3.4 to differentiate various combinations of trigonometric and other functions. EXAMPLE 1 The product rule yields Dx (x 2 sin x) = (Dx x 2 )(sin x) + (x 2 )(Dx sin x) = 2x sin x + x 2 cos x. 170 ◗ Derivatives of Trigonometric Functions SECTION 3.7 171 EXAMPLE 2 If y = cos x , then the quotient rule yields 1 − sin x dy (Dx cos x)(1 − sin x) − (cos x)[Dx (1 − sin x)] = dx (1 − sin x)2 (− sin x)(1 − sin x) − (cos x)(− cos x) = (1 − sin x)2 − sin x + sin2 x + cos2 x − sin x + 1 = = ; (1 − sin x)2 (1 − sin x)2 1 dy = . dx 1 − sin x ◗ EXAMPLE 3 If x = cos3 t and u = cos t—so that x = u 3 —then the chain rule yields d x du dx = = (3u 2 )(− sin t) = (3 cos2 t)(− sin t) = −3 cos2 t sin t. dt du dt ◗ EXAMPLE 4 If g(t) = (2 − 3 cos t)3/2 , then the chain rule yields g (t) = 32 (2 − 3 cos t)1/2 Dt (2 − 3 cos t) = 32 (2 − 3 cos t)1/2 (3 sin t) = 92 (2 − 3 cos t)1/2 sin t. ◗ EXAMPLE 5 Write an equation of the line tangent to the curve y = cos2 x at the point P on the graph where x = 0.5. Approximations are allowed. Solution The y-coordinate of P is y(0.5) = (cos 0.5)2 ≈ (0.8776)2 ≈ 0.7702. Because dy = −2 cos x sin x, dx the slope of the tangent line at P is dy m= = −2(cos 0.5)(sin 0.5) ≈ −0.8415. d x x=0.5 2 y = cos 2 x 1 y P (0.5, 0.7702) 0 −1 y = − 0.8415x + 1.1909 −2 −1 0 1 x 2 Then the point-slope formula gives the (approximate) equation 3 FIGURE 3.7.2 The curve y = cos2 x and its tangent line at the point P where x = 0.5. y − 0.7702 = −(0.8415)(x − 0.5); that is, y = −(0.8415)x + 1.1909, as the desired equation of the tangent line at P. Figure 3.7.2 shows the result of checking this computation by graphing both the curve ◗ y = cos2 x and the line with this equation. The Remaining Trigonometric Functions It is easy to differentiate the other four trigonometric functions, because they can be expressed in terms of the sine and cosine functions: tan x sec x = = sin x , cos x 1 , cos x cot x csc x = = cos x , sin x 1 . sin x (6) Each of these formulas is valid except where a zero denominator is encountered. Thus tan x and sec x are undefined when x is an odd integral multiple of π/2, and cot x and csc x are undefined when x is an integral multiple of π . The graphs of the six trigonometric functions appear in Fig. 3.7.3. There we show the sine and its reciprocal, the cosecant, in the same coordinate plane; we also pair the cosine with the secant but show the tangent and cotangent functions separately. 171 172 CHAPTER 3 The Derivative y 4 3 2 −π 2 1 y = csc x 3π 2 π 2 −1 −2 −3 −4 π y = sin x 2π 3π tan x = x sin x , cos x so (Dx sin x)(cos x) − (sin x)(Dx cos x) (cos x)2 (cos x)(cos x) − (sin x)(− sin x) cos2 x + sin2 x 1 = = = ; 2 2 cos x cos x cos2 x Dx tan x = sec2 x. Dx tan x = y y = sec x 4 3 2 −π The functions in Eq. (6) can be differentiated by using the quotient rule and the derivatives of the sine and cosine functions. For example, y = cos x π −1 −2 −3 −4 2π 3π x As an exercise (Problem 71), you should derive in similar fashion the differentiation formulas in Eqs. (8) through (10) of Theorem 2. (a) THEOREM 2 Derivatives of Trigonometric Functions The functions f (x) = tan x, g(x) = cot x, p(x) = sec x, and q(x) = csc x are differentiable wherever they are defined, and y 4 3 2 1 −π π 2 −π 2 −1 −2 −3 −4 π 3 π 2π 2 3π Dx tan x = sec2 x, x (7) Dx cot x = − csc x, Dx sec x = sec x tan x, Dx csc x = − csc x cot x. 2 y = tan x (8) (9) (10) y 4 3 2 1 −π y = cot x −1 −2 −3 −4 π 2π 3π (b) FIGURE 3.7.3 Graphs of the six trigonometric functions. x The patterns in the formulas of Theorem 2 and in Eqs. (4) and (5) make them easy to remember. The formulas in Eqs. (5), (8), and (10) are the “cofunction analogues” of those in Eqs. (4), (7), and (9), respectively. Note that the derivative formulas for the three cofunctions are those involving minus signs. EXAMPLE 6 Dx (x tan x) = (Dx x)(tan x) + (x)(Dx tan x) = (1)(tan x) + (x)(sec2 x) = tan x + x sec2 x. Dt (cot3 t) = Dt (cot t)3 = 3(cot t)2 Dt cot t = 3(cot t)2 (−csc2 t) = −3 csc2 t cot2 t. √ √ (Dz sec z) z − (sec z) Dz z sec z = Dz √ √ 2 z z √ (sec z)(tan z) z − (sec z) 12 z −1/2 = z 1 −3/2 = 2z (2z tan z − 1) sec z. ◗ Chain Rule Formulas Recall from Eq. (7) in Section 3.3 that the chain rule gives Dx [g(u)] = g (u) du dx (11) for the derivative of the composition g(u(x)) of two differentiable functions g and u. This formula yields a chain rule version of each new differentiation formula that we learn. 172 Derivatives of Trigonometric Functions SECTION 3.7 173 If we apply Eq. (11) first with g(u) = sin u, then with g(u) = cos u, and so on, we get the chain rule versions of the trigonometric differentiation formulas: Dx sin u = (cos u) du , dx Dx cos u = (− sin u) Dx tan u = (sec2 u) (12) du , dx (13) du , dx Dx cot u = (− csc2 u) (14) du , dx Dx sec u = (sec u tan u) du , dx Dx csc u = (− csc u cot u) du . dx (15) (16) (17) The cases in which u = kx (where k is a constant) are worth mentioning. For example, Dx sin kx = k cos kx and Dx cos kx = −k sin kx. (18) The formulas in (18) provide an explanation of why radian measure is more appropriate than degree measure. Because it follows from Eq. (1) that an angle of degree measure x has radian measure π x/180, the “sine of an angle of x degrees” is a new and different function with the formula πx sin x ◦ = sin , 180 expressed on the right-hand side in terms of the standard (radian-measure) sine function. Hence the first formula in (18) yields πx π cos , Dx sin x ◦ = 180 180 so Dx sin x ◦ ≈ (0.01745) cos x ◦ . The necessity of using the approximate value 0.01745 here—and indeed its very presence—is one reason why radians instead of degrees are used in the calculus of trigonometric functions: When we work with radians, we don’t need such approximations. EXAMPLE 7 If y = 2 sin 10t + 3 cos π t, then dy = 20 cos 10t − 3π sin πt. dt ◗ EXAMPLE 8 Dx (sin2 3x cos4 5x) = [Dx (sin 3x)2 ](cos4 5x) + (sin2 3x)[Dx (cos 5x)4 ] = 2(sin 3x)(Dx sin 3x) · (cos4 5x) + (sin2 3x) · 4(cos 5x)3 (Dx cos 5x) = 2(sin 3x)(3 cos 3x)(cos4 5x) + (sin2 3x)(4 cos3 5x)(−5 sin 5x) = 6 sin 3x cos 3x cos4 5x − 20 sin2 3x sin 5x cos3 5x. ◗ 173 174 CHAPTER 3 The Derivative √ EXAMPLE 9 Differentiate f (x) = cos x. √ √ Solution If u = x, then du/d x = 1/(2 x), so Eq. (13) yields Dx cos 1 y = cos y x 0 y = − 2 sin x x −1 0 20 40 60 80 x FIGURE √ 3.7.4 The curve y = cos x and the multiple √ constant √ y = −(2 sin x)/ x of its derivative. 100 √ du dx √ √ sin x 1 = − sin x √ = − √ . 2 x 2 x x = Dx cos u = (− sin u) Alternatively, we can carry out this computation without introducing the auxiliary variable u: √ √ √ √ sin x Dx cos x = − sin x · Dx x = − √ . 2 x √ In Fig. 3.7.4 we have plotted both the curve y = y(x) = cos x and (to show the vertical scale more clearly) the constant multiple √ 2 sin x y = 4y (x) = − √ x of its derivative. Note the correspondence in this figure between the local maxima and √ minima of the function y(x) = cos x and the zeros of its derivative y (x) (which are ◗ the same as the zeros of 4y (x)). EXAMPLE 10 Differentiate 2 y = sin2 (2x − 1)3/2 = sin(2x − 1)3/2 . Solution Here, y = u 2 , where u = sin(2x − 1)3/2 , so dy du = 2u = 2 sin(2x − 1)3/2 · Dx sin(2x − 1)3/2 dx dx = 2 sin(2x − 1)3/2 cos(2x − 1)3/2 · Dx (2x − 1)3/2 = 2 sin(2x − 1)3/2 cos(2x − 1)3/2 32 (2x − 1)1/2 · 2 = 6(2x − 1)1/2 sin(2x − 1)3/2 cos(2x − 1)3/2 . ◗ EXAMPLE 11 Dx tan 2x 3 = (sec2 2x 3 ) · Dx (2x 3 ) = 6x 2 sec2 2x 3 . Dt cot3 2t = Dt (cot 2t)3 = 3(cot 2t)2 · Dt (cot 2t) = (3 cot2 2t)(− csc2 2t) · Dt (2t) = −6 csc2 2t cot2 2t. √ √ sec y tan y √ √ √ √ . D y sec y = sec y tan y · D y y = √ 2 y √ Dz csc z = Dz (csc z)1/2 = 12 (csc z)−1/2 · Dz (csc z) √ = 12 (csc z)−1/2 (−csc z cot z) = − 12 (cot z) csc z. ◗ Examples 12 and 13 illustrate the applications of trigonometric functions to rateof-change and maximum-minimum problems. EXAMPLE 12 A rocket is launched vertically and is tracked by a radar station located on the ground 5 mi from the launch pad. Suppose that the elevation angle θ of the line of sight to the rocket is increasing at 3◦ per second when θ = 60◦ . What is the velocity of the rocket at this instant? 174 Derivatives of Trigonometric Functions SECTION 3.7 175 Solution First we convert the given data from degrees into radians. Because there are π/180 rad in 1◦ , the rate of increase of θ becomes π 3π = 180 60 (rad/s) at the instant when π 60π = (rad). 180 3 From Fig. 3.7.5 we see that the height y (in miles) of the rocket is θ= y = 5 tan θ. Hence its velocity is dy dθ dθ dy = · = 5(sec2 θ) . dt dθ dt dt Because sec(π/3) = 2 (Fig. 3.7.6), the velocity of the rocket is π π dy = 5 · 22 · = dt 60 3 (mi/s), about 3770 mi/h, at the instant when θ = 60◦ . ◗ 2 y θ 3 π 3 5 mi 1 FIGURE 3.7.5 Tracking an ascending rocket (Example 12). FIGURE 3.7.6 π sec = 2 (Example 12). 3 EXAMPLE 13 A rectangle is inscribed in a semicircle of radius R (Fig. 3.7.7). What is the maximum possible area of such a rectangle? R y θ x FIGURE 3.7.7 The rectangle of Example 13. Solution If we denote the length of half the base of the rectangle by x and its height by y, then its area is A = 2x y. We see in Fig. 3.7.7 that the right triangle has hypotenuse R, the radius of the circle. So x = R cos θ and y = R sin θ. (19) Each value of θ between 0 and π/2 corresponds to a possible inscribed rectangle. The values θ = 0 and θ = π/2 will yield degenerate rectangles. We substitute the data in Eq. (19) into the formula A = 2x y to obtain the area A = A(θ ) = 2(R cos θ )(R sin θ) = 2R 2 cos θ sin θ (20) as a function of θ on the closed interval [0, π/2]. To find the critical points, we differentiate: dA = 2R 2 (− sin θ sin θ + cos θ cos θ ) = 2R 2 (cos2 θ − sin2 θ). dθ 175 176 CHAPTER 3 The Derivative Because d A/dθ always exists, we have critical points only if cos2 θ − sin2 θ = 0; sin2 θ = cos2 θ; tan2 θ = 1; tan θ = ±1. The only value of θ in [0, π/2] such that tan θ = ±1 is θ = π/4. Upon evaluation of A(θ ) at each of the possible values θ = 0, θ = π/4, and θ = π/2 (the endpoints and the critical point), we find that A(0) = 0, 1 1 π 2 = 2R √ = R 2, A √ 4 2 2 π = 0. A 2 ←− absolute maximum √ 2 Thus the largest √ inscribed rectangle has area R , and its dimensions are 2x = R 2 ◗ and y = R/ 2. 3.7 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. 2. 3. 4. 5. 6. 7. 8. Dx (sin x) = cos x. If g(x) = cos x, then g (x) = sin x. Dx (x 2 sin x) = 2x sin x + x 2 cos x. If g(t) = (2 − 3 cos t)3/2 , then g (t) = 32 (2 − 3 cos t)1/2 . dy If y = y(x) = tan x, then = sec2 x. dx The notation sec2 x means sec(x 2 ). Dx (sec x) = sec x tan x. If u = u(x) is differentiable, then Dx [sec(u(x))] = [sec(u(x))] · [tan(u(x))] · u (x). 9. If A(θ ) = 2 cos θ sin θ on the interval I = [0, π], then A has a global maximum value on I . 10. An easy way to show that f (x) = sin x is continuous for all x is to observe that f (x) = cos x exists for all x. 3.7 CONCEPTS: QUESTIONS AND DISCUSSION 1. The function f is said to be even if f (−x) = f (x) for all x, odd if f (−x) = − f (x) for all x. For instance, the power function f (x) = x n is even if n is an even integer, but is odd if n is an odd integer. How can you determine if a function is even or odd by looking at its graph? Which of the six trigonometric functions are even and which are odd? 2. Give an example of a function (with domain the set of all real numbers) that is neither even nor odd. Find every function that is both even and odd. 3. The six trigonometric functions all have period 2π, meaning that f (x + 2π) = f (x) for all x. Which of the trigonometric functions have period π ? Determine the value of the constant k if the function f (t) = A cos kt + B sin kt models: 176 Derivatives of Trigonometric Functions SECTION 3.7 177 • The height of the tide at a certain beachfront location; with t in hours, the values of f (t) repeat periodically every 12 h 25 min. The average monthly rainfall in a certain locale; with t in months, the values of f (t) repeat periodically every 12 months. The average daily temperature in a certain locale; with t in days, the values of f (t) repeat periodically every 365 days. • • 4. Considering the trigonometric functions sin x, tan x, sec x, and their cofunctions, what is the pattern of signs of their derivatives? State a single short sentence telling which of the six derivative formulas include minus signs and which do not. 3.7 PROBLEMS Differentiate the functions given in Problems 1 through 20. 1. f (x) = 3 sin2 x 3. f (x) = x cos x sin x 5. f (x) = x 2. f (x) = 2 cos4 x √ 4. f (x) = x sin x cos x 6. f (x) = √ x 7. f (x) = sin x cos2 x 8. f (x) = cos3 x sin2 x 10. g(t) = (2 − cos2 t)3 9. g(t) = (1 + sin t)4 1 sin t 11. g(t) = 12. g(t) = sin t + cos t 1 + cos t 13. f (x) = 2x sin x − 3x 2 cos x 14. f (x) = x 1/2 cos x − x −1/2 sin x 15. f (x) = cos 2x sin 3x 17. g(t) = t sin 2t 2 3 19. g(t) = (cos 3t + cos 5t) 5/2 16. f (x) = cos 5x sin 7x √ 18. g(t) = t cos3 3t 1 20. g(t) = 2 sin t + sin2 3t Find dy/d x in Problems 21 through 40. 21. y = sin2 √ cos 2x x 24. y = sin3 x 4 x 26. y = sin 3x √ 28. y = cos x 22. y = x 23. y = x 2 cos(3x 2 − 1) 25. y = sin 2x cos 3x cos 3x sin 5x 29. y = sin2 x 2 √ 31. y = sin 2 x 27. y = 33. y = x sin x 2 √ √ 35. y = x sin x √ 37. y = x(x − cos x)3 39. y = cos(sin x ) 2 30. y = cos3 x 3 √ 32. y = cos 3 3 x 1 2 34. y = x cos x 36. y = (sin x − cos x)2 √ √ 38. y = x sin x + x √ 40. y = sin(1 + sin x) 51. x = sec 5t tan 3t 52. x = sec2 t − tan2 t 53. x = t sec t csc t 54. x = t 3 tan3 t 3 55. x = sec(sin t) 56. x = cot(sec 7t) sin t sec t √ 59. x = 1 + cot 5t sec t 1 + tan t √ 60. x = csc t 58. x = 57. x = In Problems 61 through 64, write an equation of the line that is tangent to the given curve y = f (x) at the point P with the given x-coordinate. Then check the plausibility of your result by plotting both the curve and the line you found on the same screen. 61. y = x cos x; 63. y = x =π 4 πx tan ; π 4 62. y = cos2 x; x =1 64. y = 41. x = tan t 42. x = sec t 43. x = (tan t) 7 45. x = t 7 tan 5t √ √ 47. x = t sec t 1 49. x = csc 2 t πx 3 sin2 ; π 3 x =5 In Problems 65 through 68, find all points on the given curve y = f (x) where the tangent line is horizontal. 65. y = cos 2x 66. y = x − 2 sin x 67. y = sin x cos x 68. y = 1 3 sin x + 2 cos2 x 2 69. Figure 3.7.8 shows the graph y = x −2 cos x and two lines of slope 1 both tangent to this graph. Write equations of these two lines. 10 5 y 0 y = x − 2 cos x Find d x/dt in Problems 41 through 60. 7 x = π/4 7 44. x = (sec 2t)7 sec t 5 46. x = t √ √ 48. x = sec t tan t 1 50. x = cot √ t −5 −10 −10 −5 0 x 5 10 FIGURE 3.7.8 The curve y = x − 2 cos x and two tangent lines each having slope 1. 177 178 CHAPTER 3 The Derivative 70. Figure 3.7.9 shows the graph 16 + sin x 3 + sin x and its two horizontal tangent lines. Write equations of these two lines. y= 12 10 8 y 75. A rocket is launched vertically upward from a point 2 mi west of an observer on the ground. What is the speed of the rocket when the angle of elevation (from the horizontal) of the observer’s line of sight to the rocket is 50◦ and is increasing at 5◦ per second? 76. A plane flying at an altitude of 25,000 ft has a defective airspeed indicator. To determine her speed, the pilot sights a fixed point on the ground. At the moment when the angle of depression (from the horizontal) of her line of sight is 65◦ , she notes that this angle is increasing at 1.5◦ per second (Fig. 3.7.12). What is the speed of the plane? 6 4 y= 2 0 −8 −4 0 x 16 + sin x 3 + sin x 4 x θ 8 25,000 ft FIGURE 3.7.9 The curve 16 + sin x and its two y= 3 + sin x horizontal tangent lines. Ground FIGURE 3.7.12 The airplane of Problem 76. 71. Derive the differentiation formulas in Eqs. (8) through (10). 72. Use the definition of the derivative to show directly that g (x) = − sin x if g(x) = cos x. 73. If a projectile is fired from ground level with initial velocity v0 and inclination angle α and if air resistance can be ignored, then its range—the horizontal distance it travels—is 1 2 v sin α cos α 16 0 (Fig. 3.7.10). What value of α maximizes R? R= 77. An observer on the ground sights an approaching plane flying at constant speed and at an altitude of 20,000 ft. From his point of view, the plane’s angle of elevation is increasing at 0.5◦ per second when the angle is 60◦ . What is the speed of the plane? 78. Find the largest possible area A of a rectangle inscribed in the unit circle x 2 + y 2 = 1 by maximizing A as a function of the angle θ indicated in Fig. 3.7.13. y 0 (x, y) α θ Ground x R FIGURE 3.7.10 The projectile of Problem 73. x2 + y2 = 1 74. A weather balloon that is rising vertically is observed from a point on the ground 300 ft from the spot directly beneath the balloon (Fig. 3.7.11). At what rate is the balloon rising when the angle between the ground and the observer’s line of sight is 45◦ and is increasing at 1◦ per second? FIGURE 3.7.13 A rectangle inscribed in the unit circle (Problem 78). 79. A water trough is to be made from a long strip of tin 6 ft wide by bending up at an angle θ a 2-ft strip on each side (Fig. 3.7.14). What angle θ would maximize the crosssectional area, and thus the volume, of the trough? y 2 2 θ Ground θ 300 2 FIGURE 3.7.11 The weather balloon of Problem 74. 178 θ FIGURE 3.7.14 The water trough of Problem 79. Derivatives of Trigonometric Functions SECTION 3.7 179 80. A circular patch of grass of radius 20 m is surrounded by a walkway, and a light is placed atop a lamppost at the circle’s center. At what height should the light be placed to illuminate the walkway most strongly? The intensity of illumination I of a surface is given by I = (k sin θ)/D 2 , where D is the distance from the light source to the surface, θ is the angle at which light strikes the surface, and k is a positive constant. 81. Find the minimum possible volume V of a cone in which a sphere of given radius R is inscribed. Minimize V as a function of the angle θ indicated in Fig. 3.7.15. π/2 θ 1 FIGURE 3.7.17 A trapezoid inscribed in a semicircle (Problem 83). 84. A logger must cut a six-sided beam from a circular log of diameter 30 cm so that its cross section is as shown in Fig. 3.7.18. The beam is symmetrical, with only two different internal angles α and β. Show that the cross section is maximal when the cross section is a regular hexagon, with equal sides and angles (corresponding to α = β = 2π/3). Note that α + 2β = 2π . (Why?) R β θ β α 30 α β β FIGURE 3.7.15 Finding the smallest cone containing a fixed sphere (Problem 81). 82. A very long rectangular piece of paper is 20 cm wide. The bottom right-hand corner is folded along the crease shown in Fig. 3.7.16, so that the corner just touches the left-hand side of the page. How should this be done so that the crease is as short as possible? FIGURE 3.7.18 A hexagonal beam cut from a circular log (Problem 84). 85. Consider a circular arc of length s with its endpoints on the x-axis (Fig. 3.7.19). Show that the area A bounded by this arc and the x-axis is maximal when the circular arc is in the shape of a semicircle. [Suggestion: Express A in terms of the angle θ subtended by the arc at the center of the circle, as shown in Fig. 3.7.19. Show that A is maximal when θ = π.] y θ Circular arc of length s θ x Crease Radius r 20 FIGURE 3.7.16 Fold a piece of paper; make the crease of minimal length (Problem 82). 83. Find the maximum possible area A of a trapezoid inscribed in a semicircle of radius 1, as shown in Fig. 3.7.17. Begin by expressing A as a function of the angle θ shown there. θ Radius r FIGURE 3.7.19 Finding the maximum area bounded by a circular arc and its chord (Problem 85). 86. A hiker starting at a point P on a straight road wants to reach a forest cabin that is 2 km from a point Q 3 km down the road from P (Fig. 3.7.20). She can walk 8 km/h along the road but only 3 km/h through the forest. She wants to minimize the time required to reach the cabin. How far down the road should she walk before setting off through the forest straight 179 180 CHAPTER 3 The Derivative for the cabin? [Suggestion: Use the angle θ between the road and the path she takes through the forest as the independent variable.] 88. Let f (x) = Cabin Forest P x 2 sin ⎩ 0 1 x if x = 0 if x = 0 (the graph of f appears in Figs. 3.7.22 and 3.7.23). Apply the definition of the derivative to show that f is differentiable at x = 0 and that f (0) = 0. 2 θ Road ⎧ ⎨ Q 3 0.08 FIGURE 3.7.20 Finding the quickest path to the cabin in the forest (Problem 86). 87. Show that the function (graphed in Fig. 3.7.21) ⎧ 1 ⎨ if x = 0 x sin f (x) = x ⎩ 0 if x = 0 (see Example 4 in Section 2.3) is not differentiable at x = 0. [Suggestion: Show that whether z = 1 or z = −1, there are arbitrarily small values of h such that [ f (h) − f (0)]/ h = z. Then use the definition of the derivative.] 0.2 0 y − 0.04 y = x2 sin (1/x) − 0.08 − 0.4 −0.2 0 x 0.2 0.4 FIGURE 3.7.22 The graph of 1 y = x 2 sin (Problem 88). x × 10 −3 y=x y = −x 2 0.1 y 0.04 y = x2 sin (1/x) y = x2 0 y 0 − 0.1 − 0.2 − 0.2 y = x sin (1/x) −0.1 0 x 0.1 −2 0.2 FIGURE 3.7.21 The graph of 1 y = x sin near x = 0. x y = −x2 −0.04 0 x 0.04 FIGURE 3.7.23 The graph in Fig. 3.7.22 magnified (Problem 88). 3.8 EXPONENTIAL AND LOGARITHMIC FUNCTIONS Until now, we have concentrated on algebraic and trigonometric functions. Exponential and logarithmic functions complete the list of the so-called elementary functions that are most important in applications of calculus. Exponential Functions An exponential function is a function of the form f (x) = a x (1) where a > 0. Note that the exponent x is the variable here; the number a, called the base, is a constant. Thus • • 180 An exponential function f (x) = a x is a constant raised to a variable power, whereas The power function p(x) = x k is a variable raised to a constant power. Exponential and Logarithmic Functions SECTION 3.8 181 In elementary algebra a rational power of the positive real number a is defined in terms of integral roots and powers. If n is a positive integer then an = a · a · a · · · a (n factors) and a −n = 1 . an Next we learn that if r = p/q where p and q are integers (with q positive), then the rational power a r is defined by √ √ p q a p/q = a p = q a . The following laws of exponents are then established for all rational exponents r and s: a r +s a −r = ar · a s , 1 = , ar (a r )s = a r ·s , (ab)r = a r · br . (2) Moreover, recall that a0 = 1 (3) for every positive real number a. The following example illustrates the fact that applications often call for irrational exponents as well as rational exponents. EXAMPLE 1 Consider a bacteria population P(t) that begins (at time t = 0) with initial population P(0) = 1 (million) and doubles every hour thereafter. The growing population is given at 1-hour intervals as in the following table: t 1 2 3 4 5 P 2 4 8 16 32 (hours) (millions) It is evident that P(n) = 2n if n is an integer. Now let’s make the plausible assumption that the population increases by the same factor in any two time intervals of the same length—for example, if it grows by 10% in any one eight-minute interval, then it grows by 10% in any other eight-minute interval. If q is a positive integer and k denotes the factor by which the population increases during a time interval of length t = 1/q, then the population is given at successive time intervals of length 1/q as in the next table. t 1 q 2 q 3 q ··· q =1 q P k k2 k3 ··· kq = 2 (Why?) We therefore see that k = 21/q . If p is another positive integer, then during p/q hours the population P will increase p times by the factor k = 21/q , so it follows that P( p/q) = k p = (21/q ) p = 2 p/q . Thus the bacteria population after t hours is given (in millions) by P(t) = 2t if the exponent t is a rational number. But because time is not restricted to rational ◗ values alone, we surely ought to conclude that P(t) = 2t for all t 0. 181 182 CHAPTER 3 The Derivative t 2t 3.1 3.14 3.141 3.1415 3.14159 3.141592 3.1415926 ↓ π 8.5742 8.8152 8.8214 8.8244 8.8250 8.8250 8.8250 ↓ 2π Investigation But what do we mean by an expression involving an irrational exponent, √ 2 such as 2 or 2π ? To find the value of 2π , we might work with (rational) finite decimal approximations to the irrational number π = 3.1415926 · · · . For example, a calculator gives 23.1 = 231/10 = P(π ) ≈ 8.8250 r →x 1 x 2 3 4 y 5 4 • • y = a x (a = 2 here) 1 −1 1 2 3 FIGURE 3.8.3 The graph of y = a x has “holes” if only rational values of x are used. (4) a r < a r · a s−r = a r +(s−r ) = a s . • 3 −2 (r rational). Thus a r < a s whenever 0 < r < s, so the exponential function f (x) = a x with a > 1 is certainly an increasing function if only positive rational values of the exponent are involved. A graphing calculator or computer actually plots only finitely many points (x, a x ), but the curve plotted in Fig. 3.8.2 looks connected because these points are plotted too close together for the eye to distinguish them. By contrast, the graph in Fig. 3.8.3 is shown with a dotted curve to suggest that it is densely filled with tiny holes corresponding to the missing points (x, a x ) for which x is irrational. In Section 6.7 we will use calculus to show that these holes can be filled to obtain the graph of a continuous increasing function f with the following properties: FIGURE 3.8.2 The graph y = 2x . 2 (million). Indeed, when the meaning of the limit in (4) is made precise, it provides one way of defining as well as calculating values of the exponential function f (x) = a x for all x. On a calculator, the ∧ key (sometimes the y x key) is ordinarily used to calculate values of exponential functions. For instance, Fig. 3.8.2 shows the result of graphing the function defined by y = 2∧x. We see the steadily rising graph (from left to right) of a function that is positive-valued for all x. Indeed, if r and s are positive rational numbers with r < s and a > 1, then we note first that a s−r > 1 (Why?) and then that y = 2x 0 ≈ 8.5742. Because any irrational number can be approximated arbitrarily closely by rational numbers, the preceding investigation suggests that the value of a x —with irrational exponent x and a fixed base a > 0—can be regarded as a limit of the form a x = lim a r −1 31 The approximate values shown in the table in Fig. 3.8.1 indicate that the bacteria population in Example 1 after π hours is FIGURE 3.8.1 Investigating 2π . 10 9 8 7 6 y 5 4 3 2 1 0 −2 √ 2 10 x f (x) is defined for every real number x; f (r ) = a r if r is rational; and the laws of exponents in (2) hold for irrational as well as rational exponents. We therefore write f (x) = a x for all x and call f the exponential function with base a. As illustrated in Fig. 3.8.4, the exponential function f (x) = a x with a > 1 increases rapidly as x > 0 increases, and the graphs of y = a x look qualitatively similar for different values of the base a so long as a > 1. The steep rate of increase of a x for x positive and increasing is a characteristic feature of exponential functions. Figures 3.8.5 and 3.8.6 compare the graphs of the exponential function y = 2x and the quadratic function y = x 2 . Derivatives of Exponential Functions To compute the derivative of the exponential function f (x) = a x , we begin with the definition of the derivative and then use the first law of exponents in Eq. (2) to simplify. 182 Exponential and Logarithmic Functions SECTION 3.8 183 80 y a = 10 10 9 8 7 6 y 5 4 3 2 1 0 −2 a=5 a=3 40 a=2 0 0 2 4 6 x FIGURE 3.8.4 y = a x for a = 2, 3, 5, 10. 200 180 160 140 120 y 1000 80 60 40 20 0 −2 (2, 4) y = 2x y = x2 −1 0 1 x 3 2 4 y = 2x y = x2 0 2 4 6 8 x FIGURE 3.8.6 But here we see 2x increasing much more rapidly than x 2 . FIGURE 3.8.5 Here the graphs y = 2x and y = x 2 look similar for x > 2. This gives f (x + h) − f (x) a x+h − a x = lim h→0 h→0 h h ax ah − ax (by the laws of exponents) = lim h→0 h ah − 1 (because a x is “constant” with respect to h). = a x lim h→0 h f (x) = lim Under the assumption that f (x) = a x is differentiable, it follows that the limit ah − 1 h→0 h m(a) = lim (5) exists. Although its value m(a) depends on a, the limit is a constant as far as x is concerned. Thus we find that the derivative of a x is a constant multiple of a x itself: Dx a x = m(a) · a x . h 2h − 1 h 3h − 1 h 0.1 0.01 0.001 0.0001 0.718 0.696 0.693 0.693 1.161 1.105 1.099 1.099 FIGURE 3.8.7 Investigating the values of m (2) and m (3). (6) Because a 0 = 1, we see from Eq. (6) that the constant m(a) is the slope of the line tangent to the curve y = a x at the point (0, 1), where x = 0. The numerical data shown in Fig. 3.8.7 suggest that m(2) ≈ 0.693 and that m(3) ≈ 1.099. The tangent lines with these slopes are shown in Fig. 3.8.8. Thus it appears that Dx 2x ≈ (0.693) · 2x and Dx 3x ≈ (1.099) · 3x . y (7) y y = 3x y = 2x Slope ≈ 1.099 Slope ≈ 0.693 (0, 1) (0, 1) x (a) x (b) FIGURE 3.8.8 The graphs (a) y = 2x and (b) y = 3x . 183 184 CHAPTER 3 The Derivative We would like somehow to avoid awkward numerical factors like those in Eq. (7). It seems plausible that the value m(a) defined in Eq. (5) is a continuous function of a. If so, then because m(2) < 1 and m(3) > 1, the intermediate value theorem implies that m(e) = 1 (exactly) for some number e between 2 and 3. If we use this particular number e as the base, then it follows from Eq. (6) that the derivative of the resulting exponential function f (x) = e x is Dx e x = e x . (8) Thus the function e x is its own derivative. We call f (x) = e x the natural exponential function. Its graph is shown in Fig. 3.8.9. y y = ex n 1+ 1 n n Slope: 1 10 100 1,000 10,000 100,000 (0, 1) x FIGURE 3.8.9 The graph y = e x . 2.594 2.705 2.717 2.718 2.718 FIGURE 3.8.10 Numerical estimate of the number e. We will see in Section 4.9 that the number e is given by the limit 1 n . e = lim 1 + n→∞ n Let us investigate this limit numerically. With a calculator we obtain the values in the table of Fig. 3.8.10. The evidence suggests (but does not prove) that e ≈ 2.718 to three places. This number e is one of the most important special numbers in mathematics. It is known to be irrational; its value accurate to 15 places is e ≈ 2.71828 1828 459045. The chain rule version of Eq. (8) is the differentiation formula Dx eu = eu du , dx where u denotes a differentiable function of x. In particular, Dx ekx = kekx if k is a constant. For instance, Dx e−x = −e−x and Dx e2x = 2e2x . EXAMPLE 2 (a) If f (x) = x 2 e−x , then the product rule gives f (x) = (Dx x 2 )e−x + x 2 (Dx e−x ) = (2x)e−x + x 2 (−e−x ) = (2x − x 2 )e−x . 184 (9) Exponential and Logarithmic Functions SECTION 3.8 185 1 (b) If y = e2x , then the quotient rule gives 2x + 1 (Dx e2x )(2x + 1) − (e2x )Dx (2x + 1) dy = dx (2x + 1)2 (2e2x )(2x + 1) − (e2x )(2) 4xe2x = = . (2x + 1)2 (2x + 1)2 y = x2 e−x 0.5 y 0 −0.5 −1 0 1 2 3 4 5 6 7 8 x ◗ EXAMPLE 3 Figure 3.8.11 shows a computer plot of the graph of f (x) = x 2 e−x . Find the coordinates of the indicated local maximum point on the curve in the first quadrant. Solution The calculation in part (a) of Example 2 yields f (x) = 0 when FIGURE 3.8.11 The graph of Example 3. x(2 − x) = 0, ex so the only critical points of f are at x = 0 and x = 2. Thus the indicated first-quadrant ◗ critical point on the curve is (2, f (2)) = (2, 4e−2 ) ≈ (2, 0.5413). (2x − x 2 )e−x = x Logarithms and Inverse Functions In precalculus courses, the base a logarithm function loga x is introduced as the “opposite” of the exponential function f (x) = a x with base a > 1. That is, loga x is the power to which a must be raised to get x. Thus f y = loga x f(x) if and only if a y = x. (10) With a = 10, this is the base 10 common logarithm log10 x. EXAMPLE 4 g x log10 1000 = 3 because 1000 = 103 ; log10 (0.1) = −1 because 0.1 = 10−1 ; log2 16 = 4 because 16 = 24 ; log3 9 = 2 because 9 = 32 . ◗ If y = loga x, then a y = x > 0. Hence it follows that x a loga x = x 11(a) loga (a y ) = y. 11(b) and Thus the base a exponential and logarithmic functions are natural opposites, in the sense that each undoes the result of applying the other. Apply both in succession—in either order—and you’re back where you started (Fig. 3.8.12). Example 5 gives other familiar pairs of functions that are inverses of each other. g g(x) EXAMPLE 5 The following are some pairs of inverse functions: (a) f (x) = x + 1 and g(x) = x − 1. f x FIGURE 3.8.12 Inverse functions f and g. Each undoes the effect of the other. Adding 1 and subtracting 1 are inverse operations; doing either undoes the other. Next, doubling and halving are inverse operations: x (b) f (x) = 2x and g(x) = . 2 A function can be its own inverse: 1 1 and g(x) = . (c) f (x) = x x ◗ 185 186 CHAPTER 3 The Derivative Like f (x) = a x and g(x) = loga x, each pair f and g of functions given in Example 5 has the property that f (g(x)) = x and g( f (x)) = x (12) for all values of x in the domains of g and f , respectively. For instance, the functions f (x) = x + 1 and g(x) = x − 1 in part (a) of the example are defined for all x, and it is easy to check that f (g(x)) = g(x) + 1 = (x − 1) + 1 = x and g( f (x)) = f (x) − 1 = (x + 1) − 1 = x for every real number x. y = f − (x) y DEFINITION Inverse Functions The two functions f and g are inverse functions, or are inverses of each other, provided that • The range of values of each function is the domain of definition of the other, and • The relations in (12) hold for all x in the domains of g and f , respectively. y = f + (x) The following two examples illustrate the fact that care is required when we specify the domains of definition of the functions f and g to ensure that the condition in (12) is satisfied. y = x2 x=− y x=+ y x FIGURE 3.8.13 The function f (x) = x 2 and its restrictions f − and f + . EXAMPLE 6 The function f (x) = x 2 is defined for all x and its range is the set of all nonnegative real numbers y; thus we write f : (−∞, +∞) −→ [0, +∞). As indicated in Fig. 3.8.13, it is a familiar fact that each positive number y has two √ √ different square roots, g+ (y) = + y and g− (y) = − y. (Recall that the symbol √ y unadorned with either sign always denotes the nonnegative square root of y.) The √ positive square root function g+ (y) = + y is defined for all y 0, as is g− (y), and they are the inverses of the two different squaring functions and and y = x3 x= y y = x3 defined by f + (x) = x 2 for x 0 f − : (−∞, 0] −→ [0, +∞) defined by f − (x) = x 2 for x 0. (The functions f + and f − are obtained by “restricting” the function f (x) = x 2 to the nonnegative x-axis and the nonpositive x-axis, respectively.) For instance, √ √ f − (g− (x)) = (− x)2 = ( x)2 = x for all x 0 y 3 f + : [0, +∞) −→ [0, +∞) x= 3 y x FIGURE 3.8.14 The function √ f (x) = x 3 has inverse g(y) = 3 y defined for all y. √ g− ( f − (x)) = − x 2 = − (−x)2 = −(−x) = x for all Thus the functions f − and g− are inverse functions. You should verify similarly that ◗ the functions f + and g+ are inverse functions. √ EXAMPLE 7 In contrast with Example 6, the functions f (x) = x 3 and g(x) = 3 x are inverse functions defined for all x. The difference is that any real number x— whether positive, negative, or zero—has one and only one cube root (as indicated in ◗ Fig. 3.8.14). Because a x > 0 for all x (as illustrated in Fig. 3.8.15), it follows that loga x is defined only for x > 0. Because interchanging x and y in a y = x yields y = a x , it follows from Eq. (10) that the graph of y = loga x is the reflection in the line y = x of the graph of y = a x and therefore has the shape shown in Fig. 3.8.15. Because a 0 = 1, it also follows that loga 1 = 0, so the intercepts in the figure are independent of the choice of a. 186 x 0. Exponential and Logarithmic Functions SECTION 3.8 187 y y = ax 3 y=x 2 x = a y or y = loga x 1 −2 −1 1 2 3 x −1 −2 −3 FIGURE 3.8.15 The graph of x = a y is the graph of the inverse function loga x of the exponential function a x . The case a > 1 is shown here. We can use the inverse-function relationship between loga x and a x to deduce, from the laws of exponents in Eq. (2), the following laws of logarithms: loga x y x loga y = loga x + loga y, = loga x − loga y, 1 x loga x y loga = − loga x, = (13) y loga x. We will verify these laws of logarithms in Section 6.7. Derivatives of Inverse Functions Our interest in inverse-function pairs at this point stems from the following general principle: When we know the derivative of either of two inverse functions, then we can use the inverse-function relationship between them to discover the derivative of the other of the two functions. Theorem 1 is usually proved in an advanced calculus course. THEOREM 1 Differentiation of an Inverse Function Suppose that the differentiable function f is defined on the open interval I and that f (x) > 0 for all x in I . Then f has an inverse function g, the function g is differentiable, and 1 g (x) = (14) f (g(x)) for all x in the domain of g. COMMENT 1 Theorem 1 is true also when the condition f (x) > 0 is replaced with the condition f (x) < 0. If we assume that g is differentiable, then we can derive the formula in Eq. (14) by differentiating with respect to x each side in the inverse-function relation f (g(x)) = x. When we differentiate each side, using the fact that this relation is actually an identity on some interval and using the chain rule on the left-hand side, the result is f (g(x)) · g (x) = 1. When we solve this equation for g (x), the result is Eq. (14). 187 188 CHAPTER 3 The Derivative COMMENT 2 In order that the function f in Theorem 1 have an inverse function g, it is necessary (and sufficient) that, for each y in the range of f , there exists exactly one x in the domain of f such that f (x) = y. (We can then define g(y) = x.) Figure 3.8.14 indicates that this is so for the cubing function f (x) = x 3 of Example 7. In contrast, we see in Fig. 3.8.13 that each y > 0 in the range of the squaring function f (x) = x 2 of Example 6 corresponds to two different values of x—the positive and negative square roots of y. This is why the squaring function f : (−∞, +∞) −→ [0, +∞) has no (single) inverse function. The graph of f is the entire parabola in the figure. The right and left “halves” of the parabola are the graphs of the restrictions f + and f − with inverse functions g+ and g− , respectively. COMMENT 3 Equation (14) is easy to remember in differential notation. Let us write x = f (y) and y = g(x). Then dy/d x = g (x) and d x/dy = f (y). So Eq. (14) becomes the seemingly inevitable formula dy 1 = . dx dx dy (15) In using Eq. (15), it is important to remember that dy/d x is to be evaluated at x, but d x/dy is to be evaluated at the corresponding value of y; namely, y = g(x). EXAMPLE 8 In Section 3.4 we verified the power rule Dx x r = r x r −1 for rational values of the exponent r . But there we needed to know in advance that Dx x 1/q = (1/q)x (1/q)−1 for every positive integer q. Now we observe that the power function f (x) = x q , x > 0 certainly has a positive derivative: f (x) = q x q−1 √ for x > 0. Therefore Theorem 1 implies that its inverse function g(x) = x 1/q = q x exists and has derivative 1 1 1 1 = = x (1/q)−1 , = Dx x 1/q = g (x) = q−1 1−(1/q) f (g(x)) qx q q x 1/q as desired. Alternatively, we could use the approach of Comment 1 and simply write the identity (x 1/q )q = x. Then differentiation, using the chain rule on the left (and Dx x ≡ 1 on the right) gives the equation q(x 1/q )q−1 · Dx x 1/q = 1, so we can solve for ◗ Dx x 1/q . The Natural Logarithm The natural exponential function f (x) = e x is defined for all x and f (x) = e x > 0. If f is the inverse function that consequently is guaranteed by Theorem 1 in this section, then f (g(x)) = e g(x) = x. Thus g(x) is “the power to which e must be raised to get x,” and therefore is simply the logarithm function with base e: g(x) = loge x. The function g is therefore called the natural logarithm function. It is commonly denoted (on calculator keys, for instance) by the special symbol ln: 4 3 ln x = loge x y = ln x 2 −1 1 e −2 −3 −4 0 2 4 6 8 eln x = x 10 x FIGURE 3.8.16 The graph of the natural logarithm function. for all x > 0 (17a) and ln(e x ) = x 188 (16) Because e x > 0 for all x, it follows that ln x is defined only for x > 0. The graph of y = ln x is shown in Fig. 3.8.16, and appears to rise quite slowly when x is large. We note that ln 1 = 0, so the graph has x-intercept x = 1, and that ln e = 1 (because ln e = loge e = 1). The inverse function relations between f (x) = e x and g(x) = ln x are these: 1 y 0 (x > 0). for all x. (17b) Exponential and Logarithmic Functions SECTION 3.8 189 Derivatives of Logarithmic Functions To differentiate the natural logarithm function, we can apply Eq. (14) in Theorem 1 (with f (x) = e x and g(x) = ln x) and thereby write g (x) = 1 f (g(x)) = 1 f (ln x) = 1 eln x = 1 . x Alternatively, we could begin with Eq. (17a) and differentiate both sides with respect to x, as follows: Dx eln x = Dx x; eln x · Dx ln x = 1 x · Dx ln x = 1. (by Eq. (9) with u = ln x); Thus we find either way that the derivative g (x) = Dx ln x of the natural logarithm function is given by Dx ln x = 1 x (18) for x > 0. Thus ln x is the hitherto missing function whose derivative is x −1 = 1/x. Just as with exponentials, the derivative of a logarithm function with base other than e involves an inconvenient numerical factor. For instance, Problem 74 shows that Dx log10 x ≈ 0.4343 . x (19) The contrast between Eqs. (18) and (19) illustrates one way in which base e logarithms are “natural.” EXAMPLE 9 Figure 3.8.17 shows the graph of the function 1 0.5 y f (x) = y = ln x x ln x . x Find the coordinates of the indicated first-quadrant critical point on this curve. 0 Solution Equation (18) and the quotient rule yield -0.5 -1 0 5 15 10 20 x FIGURE 3.8.17 The graph of Example 9. 25 1 · x − (ln x) · 1 (D 1 − ln x ln x)(x) − (ln x)(D x) x x x = = . f (x) = 2 2 x x x2 Hence the only critical point of f occurs when ln x = 1; that is, when x = e. Thus the ◗ critical point indicated in Fig. 3.8.17 is (e, 1/e) ≈ (2.718, 0.368). The chain-rule version of Eq. (18) is Dx ln u = u 1 du · = , u dx u (20) where u is a positive-valued function of x and u denotes u (x). If u(x) has negative values, then the function ln |u| is defined wherever u is nonzero, so Dx ln |u| = 1 d|u| du · · |u| du d x by Eq. (20) with |u| in place of u. But from the familiar graph of the absolute value function we see that |u| d|u| −1 if u < 0, = = +1 if u > 0. du u 189 190 CHAPTER 3 The Derivative It therefore follows that y y = ln |x | 3 y= 2 1 x Dx ln |u| = 1 du · u dx (21) 1 −6 −4 −2 2 4 6 x wherever the differentiable function u(x) is nonzero. In particular, −1 Dx ln |x| = −2 −3 FIGURE 3.8.18 The function f (x) = ln |x| and its derivative f (x) = 1/x. 1 x (22) if x = 0 (see Fig. 3.8.18). EXAMPLE 10 With u = 1 + x 2 as the “inner function” in Eq. (20), we get 2x u = . u 1 + x2 √ EXAMPLE 11 Find the derivative of y = 1 + ln x. Dx ln(1 + x 2 ) = ◗ Solution Now u = 1 + ln x is the inner function, so 1 dy = (1 + ln x)−1/2 · Dx (1 + ln x) dx 2 1 1 1 . = (1 + ln x)−1/2 · = √ 2 x 2x 1 + ln x 2x + 3 . EXAMPLE 12 Find the derivative of y = ln 4x + 5 ◗ Solution If we differentiated immediately, we’d find ourselves applying the quotient rule to differentiate the fraction within the radical. (Try it yourself!) It’s simpler to apply laws of logarithms to simplify the given function before differentiating it: 1 2x + 3 1/2 1 2x + 3 = [ln(2x + 3) − ln(4x + 5)]. = ln y = ln 4x + 5 2 4x + 5 2 Then 1 2 4 1 2 1 dy = − = − =− 2 . dx 2 2x + 3 4x + 5 2x + 3 4x + 5 8x + 22x + 15 ◗ Logarithmic Differentiation The derivatives of certain functions are most conveniently found by first differentiating their logarithms. This process—called logarithmic differentiation—involves the following steps for finding f (x). 1. Given: 2. Take natural logarithms; then simplify, using laws of logarithms: 3. Differentiate with respect to x: 4. Multiply both sides by y = f (x): y = f (x) ln y = ln f (x) 1 dy · = Dx [ln f (x)] y dx dy = f (x)Dx [ln f (x)] dx REMARK If f (x) is not positive-valued everywhere, then Steps 1 and 2 should be replaced with y = | f (x)| and ln y = ln | f (x)|, respectively. The differentiation in Step 3 then leads to the result dy/d x = f (x)Dx [ln | f (x)|] in Step 4. In practice, we need not be overly concerned in advance with the sign of f (x), because the appearance of what seems to be the logarithm of a negative quantity will signal the fact that absolute values should be used. 190 Exponential and Logarithmic Functions SECTION 3.8 191 EXAMPLE 13 Find dy/d x, given (x 2 + 1)3 y= . 3 (x 3 + 1)4 Solution The laws of logarithms give ln y = ln 3 4 (x 2 + 1)3/2 = ln(x 2 + 1) − ln(x 3 + 1). 3 4/3 (x + 1) 2 3 Then differentiation with respect to x gives 1 dy 3 2x 4 3x 2 3x 4x 2 · = · 2 − · 3 = 2 − 3 . y dx 2 x +1 3 x +1 x +1 x +1 Finally, to solve for dy/d x, we multiply both sides by y= (x 2 + 1)3/2 , (x 3 + 1)4/3 and we obtain dy = dx 3x 4x 2 − x2 + 1 x3 + 1 · (x 2 + 1)3/2 . (x 3 + 1)4/3 ◗ EXAMPLE 14 Find dy/d x, given y = x x+1 for x > 0. Solution If y = x x+1 , then ln y = ln(x x+1 ) = (x + 1) ln x; 1 dy 1 1 · = (1)(ln x) + (x + 1) = 1 + + ln x. y dx x x Multiplying by y = x x+1 gives 1 dy = 1 + + ln x x x+1 . dx x ◗ 3.8 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. An exponential function has the form f (x) = a x where a is a constant. If r and s are rational numbers and a > 0, then (a r )s = a r +s . If a > 0 then f (x) = a x is an increasing function. If a > 0 and f (x) = a x , then f (x) = xa x−1 . Dx (e x ) = e x . Dx (x 2 e−x ) = 2xe−x − x 2 e−x . A function can be its own inverse. If a > 1, then y = loga x if and only if x = a y . 1 9. Dx (ln x) = . x ln x has coordinates (e, e−1 ). 10. The highest point on the graph of f (x) = x 1. 2. 3. 4. 5. 6. 7. 8. 191 192 CHAPTER 3 The Derivative 3.8 CONCEPTS: QUESTIONS AND DISCUSSION 1. Example 5 lists three inverse function pairs. List several more inverse function pairs f and g of your own. In each case specify the domains of f and g and verify that f and g are indeed inverse functions. 2. Suppose that n is a positive integer. Discuss (as in Examples 6 and 7) the question √ of whether the power function f (x) = x n and the root function g(x) = n x are inverse functions defined for all x. How does the situation depend on whether n is even or odd? Discuss positive and negative nth roots if necessary. Specify the domain of definition of each function you mention and verify all claims you make. 3. Sketch the bell-shaped graph of the function 1 . 1 + x2 Explain why f (which is defined for all x) does not have an inverse function, but its restrictions f + and f − to the positive and negative x-axes do have inverse functions g+ and g− (using notation similar to that in Example 6). Find formulas for g+ (x) and g− (x) and specify the domain of definition of each of these two inverse functions. 4. Restrict the domain of each of the six trigonometric functions sin x, cos x, tan x, cot x, sec x, and csc x to those points in the interval 0 < x < π at which they are defined. Consulting the graphs in Fig. 3.7.3 as necessary, determine which of these functions have inverse functions. Answer the same question if instead the domains are restricted to the interval 0 < x < π/2. f (x) = 3.8 PROBLEMS Differentiate the functions in Problems 1 through 38. 1. f (x) = e 2x 2. f (x) = e 3. f (x) = e x2 4. f (x) = e4−x 5. f (x) = e1/x 7. g(t) = te √ 2 t 11. g(t) = ecos t x 8. g(t) = (e2t + e3t )7 √ 10. g(t) = et − e−t 14. f (x) = e−1/x √ √ 16. f (x) = e x + e− x √ 18. f (x) = e2x + e−2x 19. f (x) = sin(2e x ) 20. f (x) = cos(e x + e−x ) 21. f (x) = ln(3x − 1) √ 23. f (x) = ln 1 + 2x √ 25. f (x) = ln 3 x 3 − x 22. f (x) = ln(4 − x 2 ) 24. f (x) = ln[(1 + x)2 ] 26. f (x) = ln(sin2 x) In Problems 39 through 46, apply laws of logarithms to simplify the given function before finding its derivative. 1−x 39. f (x) = ln[(2x + 1)3 (x 2 − 4)4 ] 40. f (x) = ln 1+x √ 4 − x2 4x − 7 42. f (x) = ln 41. f (x) = ln 2 9+x (3x − 2)3 x +1 1 43. f (x) = ln 44. f (x) = x 2 ln x −1 2x + 1 √ x +1 t2 46. f (x) = ln 45. g(t) = ln 2 t +1 (x − 1)3 In Problems 47 through 58, find dy/d x by logarithmic differentiation. 47. y = 2x 49. y = x ln x √ 51. y = (ln x) x (1 + x 2 )3/2 53. y = (1 + x 3 )4/3 48. y = x x 50. y = (1 + x)1/x 52. y = (3 + 2x )x 54. y = (x + 1)x 1 x 56. y = 1 + x 27. f (x) = cos(ln x) 1 29. f (x) = ln x √ 31. f (x) = ln x x 2 + 1 28. f (x) = (ln x)3 33. f (x) = ln(cos x) 34. f (x) = ln(2 sin x) In Problems 59 through 62, write an equation of the line tangent to the graph of the given function at the indicated point. 35. f (t) = t 2 ln(cos t) 36. f (x) = sin(ln 2x) √ 38. g(t) = t[cos(ln t)]2 59. y = xe2x at the point (1, e2 ) 60. y = e2x cos x at the point (0, 1) 37. g(t) = t (ln t) 192 3 12. f (x) = xesin x −t 1−e t 1−x 15. f (x) = ex 17. f (x) = ee 3 6. f (x) = x 2 e x 9. g(t) = (t 2 − 1)e−t 13. g(t) = 3x−1 2 30. f (x) = ln(ln x) 32. g(t) = t 3/2 ln(t + 1) 55. y = (x 2 + 1)x 57. y = √ √ x x 2 58. y = x sin x 61. y = x 3 ln x at the point (1, 0) ln x 62. y = 2 at the point (e, e−2 ) x In Problems 63 and 64, differentiate the given function f (x) and its derivative in turn, several times in succession. Then give a likely formula for the result after n successive differentiations in this manner. 63. f (x) = e2x 64. f (x) = xe x 65. Figure 3.8.19 shows the graph of the function f (x) = e−x/6 sin x, together with the graphs of its “envelope curves” y = e−x/6 and y = −e−x/6 . Find the first local maximum point and the first local minimum point on the graph of f for x > 0. y 0 y = −e−x/6 −1 0 5 10 that “converges” much more rapidly. Using a calculator or computer, substitute k = 1, 2, 3, . . . , 8 in turn to discover that e ≈ 2.71828 accurate to five decimal places. 70. Suppose that u and v are differentiable functions of x. Show by logarithmic differentiation that Dx (u v ) = v(u v−1 ) du dv + (u v ln u) . dx dx Interpret the two terms on the right in relation to the special cases in which (a) u is a constant; (b) v is a constant. 71. Suppose that y = uvw/ pqr , where u, v, w, p, q, and r are nonzero differentiable functions of x. Show by logarithmic differentiation that y = e−x/6 sin x y = e−x/6 1 Exponential and Logarithmic Functions SECTION 3.8 193 1 n 69. If we substitute n = 10k in e = lim 1 + , we get the n→∞ n limit k 1 10 e = lim 1 + k k→∞ 10 15 x FIGURE 3.8.19 The graph for Problems 65 and 66. 66. Find the first two points of tangency of the curve y = e−x/6 sin x with the two envelope curves shown in Fig. 3.8.19. Are these the same as the two local extreme points found in Problem 65? 67. Find graphically the coordinates (accurate to three decimal places) of the intersection point of the graphs y = e x and y = x 10 indicated in Fig. 3.8.20. dy = dx 1 du 1 dv 1 dw 1 dp 1 dq 1 dr y· + + − − − . u dx v dx w dx p dx q dx r dx Is the generalization—for an arbitrary finite number of factors in numerator and denominator—obvious? 72. Show that the number log2 3 is irrational. [Suggestion: Assume to the contrary that log2 3 = p/q where p and q are positive integers; then express the consequence of this assumption in exponential form. Under what circumstances can an integral power of 2 equal an integral power of 3?] 73. The log key on the typical calculator denotes the base 10 logarithm f (x) = log10 x. (a) Use the definition of the derivative to show that f (1) = lim log10 (1 + h)1/ h . 10 9 8 7 6 y 5 4 3 2 1 0 h→0 y = x10 (b) Investigate the limit in (a) numerically to show that f (1) ≈ 0.4343. 74. The object of this problem is to differentiate the base 10 logarithm function of Problem 73. (a) First use the known formula for Dx eu to show that Dx 10x = 10x ln 10. (b) Conclude from the chain rule that y = ex 0 0.5 1 x 1.5 2 FIGURE 3.8.20 Comparing y = e x and y = x 10 . 68. See Problem 67. Determine a viewing rectangle that reveals a second intersection point (with x > 10) of the graphs y = e x and y = x 10 . Then determine graphically the first three digits of the larger solution x of the equation e x = x 10 (thus writing this solution in the form p.qr × 10k ). Dx 10u = 10u (ln 10) du . dx (c) Substitute u = log10 x in the inverse function identity 10log10 x = x and then differentiate using the result of part (b) to conclude that Dx log10 x = 0.4343 1 ≈ , x ln 10 x consistent with the result (for x = 1) of Problem 73. 193 194 CHAPTER 3 The Derivative 3.8 INVESTIGATION: Discovering the Number e for Yourself You can investigate the value of e by approximating the value of a such that a m(a) 2 2.7 ↓ 0.6931 0.9933 ↓ e ↑ 1.0000 ↑ 2.8 3 1.0296 1.0986 ah − 1 = 1. h→0 h You need use only available technology to calculate (with appropriate fixed values of a) values of the function φ(h) = (a h − 1)/ h with h sufficiently small that you can recognize (to appropriate accuracy) the value of the limit. For instance, if you calculate φ(h) with a = 2 and with a = 3 for h = 0.1, 0.01, 0.001, 0.0001, . . . , you should find that m(a) = lim m(2) ≈ 0.6931 < 1 whereas m(3) ≈ 1.0986 > 1. It follows that the mysterious number e for which m(e) = 1 is somewhere between 2 and 3. Linear interpolation between the values of m(2) ≈ 0.6931 and m(3) ≈ 1.0986 suggests that e ≈ 2.7 or e ≈ 2.8 accurate to one decimal place. Investigate the values of m(2.7) and m(2.8) to verify the entries shown in Fig. 3.8.21. Continue in this way to close in on the number e. Don’t quit until you’re convinced that e ≈ 2.718 accurate to three decimal places. FIGURE 3.8.21 Closing in on the number e. 3.9 IMPLICIT DIFFERENTIATION AND RELATED RATES A formula such as y = x 3 sin x defines y “explicitly” as a function of x. Most of the functions we have seen so far have been defined explicitly in this way. Nevertheless, a function can also be defined “implicitly” by an equation that can be solved for y in terms of x. Indeed, we will see that a single equation relating the two variables x and y can implicitly define two or more different functions of x. y y=+ x x EXAMPLE 1 (a) When we solve the equation y=− x x − y2 = 0 √ for y = ± x, we get the two explicit functions √ √ f (x) = x and g(x) = − x FIGURE 3.9.1 The equation x − y 2 = 0 implicitly defines the √ two functions f (x) = x and √ g(x) = − x. y y = + 100 − x 2 x y = − 100 − x 2 FIGURE 3.9.2 The equation x 2 + y 2 = 100 implicitly defines the two functions √ f (x) = 100 − x 2 and √ g(x) = − 100 − x 2 . 194 that we say are implicitly defined by the original equation. The graphs of these two functions—both defined for x 0—are the upper and lower branches of the parabola shown (in different colors) in Fig. 3.9.1. The whole parabola is the graph of the equation x − y 2 = 0 (or x = y 2 ) but is not the graph of any single function. (Why?) (b) Similarly, the equation x 2 + y 2 = 100 implicitly defines the two continuous functions f (x) = 100 − x 2 and g(x) = − 100 − x 2 —both defined for −10 x 10—that correspond to the solutions y = √ ± 100 − x 2 for y in terms of x. The graphs of f and g are the upper and lower semicircles of the whole circle x 2 + y 2 = 100 (shown in different colors ◗ in Fig. 3.9.2). Whereas the equations x − y 2 = 0 and x 2 + y 2 = 100 are readily solved for y in terms of x, an equation such as x 3 + y 3 = 3x y or sin(x + 2y) = 2x cos y may be difficult or impossible to solve for an implicitly defined function y(x). And yet the derivative dy/d x can be calculated without first expressing y in terms of x. Here’s how: We can use the chain rule and other basic differentiation rules to differentiate both sides of the given equation with respect to x (we think of x as the independent variable, although it is permissible to reverse the roles of x and y). We then solve the resulting Implicit Differentiation and Related Rates SECTION 3.9 195 equation for the derivative y (x) = dy/d x of the implicitly defined function y(x). This process is called implicit differentiation. In the examples and problems of this section, we proceed on the assumption that our implicitly defined functions actually exist and are differentiable at almost all points in their domains. (The functions with the graphs shown in Fig. 3.9.2 are not differentiable at the endpoints of their domains.) EXAMPLE 2 Use implicit differentiation to find the derivative of a differentiable function y = f (x) implicitly defined by the equation x 2 + y 2 = 100. Solution The equation x 2 + y 2 = 100 is to be regarded as an identity that implicitly defines y = y(x) as a function of x. Because x 2 + [y(x)]2 is then a function of x, it has the same derivative as the constant function 100 on the right-hand side of the identity. Thus we may differentiate both sides of the identity x 2 + y 2 = 100 with respect to x and equate the results. We obtain 2x + 2y dy = 0. dx In this step, it is essential to remember that y is a function of x, so the chain rule yields Dx (y 2 ) = 2y Dx y. Then we solve for dy x =− . dx y (1) It may be surprising to see a formula for dy/d x containing both x and y, but such a formula can be just as useful as one containing only x. For example, the formula in Eq. (1) tells us that the slope of the line tangent to the circle x 2 + y 2 = 100 at the point (6, 8) is dy 3 6 =− =− . d x (6,8) 8 4 ◗ The circle and this line are shown in Fig. 3.9.3. y Slope m = − 3 4 (6, 8) x FIGURE 3.9.3 The circle x 2 + y 2 = 100 and the tangent line at the point (6, 8). √ NOTE If we solve for y = ± 100 − x 2 in Example 1, then dy −x x = √ =− , 2 dx y ± 100 − x in agreement with Eq. (1). gives us the derivatives of √ √ Thus Eq. (1) simultaneously both the functions y = + 100 − x 2 and y = − 100 − x 2 implicitly defined by the equation x 2 + y 2 = 100. 195 196 CHAPTER 3 The Derivative EXAMPLE 3 The folium of Descartes is the graph of the equation x 3 + y 3 = 3x y y 3 x+y=3 2 ( 32 , 32 ) 1 −2 −1 x + y = −1 1 3 x 2 −1 −2 −3 FIGURE 3.9.4 A tangent line and an apparent asymptote to the curve x 3 + y 3 = 3x y. (2) This curve was first proposed by René Descartes as a challenge to Pierre de Fermat (1601–1665) to find its tangent line at an arbitrary point. The project for this section tells how we constructed Fig. 3.9.4. It indicates that the second- and fourth-quadrant points on the graph for which |x| and |y| are both large lie very close to the straight line x + y + 1 = 0. In the first quadrant we see a loop shaped like a laurel leaf—hence the name folium. (Can you see directly from Eq. (2) that the third quadrant contains no points of the folium?) Here we want to find Fermat’s answer as to the slope of a typical line tangent to the folium of Descartes. Solution Equation (2) is a cubic equation in x, and we see in Fig. 3.9.4 three different branches of the graph over an interval to the right of the origin. When we asked a computer algebra system to solve the equation for these implicitly defined functions of x, it produced three different expressions, the simplest of which was 13 2x y= −4x 3 + 4 x 6 − 4x 3 + . √ 3 2 −4x 3 + 4 x 6 − 4x 3 (It turns out that this formula describes the upper part of the loop in Fig. 3.9.4.) Surely you would not relish explicit differentiation of this expression to find the slope of a line tangent to the folium. Fortunately, the alternative of implicit differentiation is available. We need only differentiate each side of Eq. (2) with respect to x, remembering that y is a function of x. Hence we use the chain rule to differentiate y 3 and the product rule to differentiate 3x y. This yields dy dy = 3y + 3x . dx dx We can now collect coefficients (those involving dy/d x and those not) and solve for the derivative: dy (3y 2 − 3x) = 3y − 3x 2 ; dx dy y − x2 = 2 . (3) dx y −x 3x 2 + 3y 2 For instance, at the point P 3 3 , 2 2 dy dx 3 2 1 y = 1x 2 y 0 −1 of the folium, the slope of the tangent line is 3 2 2 3 2 3 − 2 2 − = −1. This result agrees with our intuition about the figure, because the evident symmetry of the folium around the line y = x suggests that the tangent line at P should, indeed, have slope −1. The equation of this tangent line is y− −2 = 3 3 , 2 2 3 2 3 2 =− x− 3 2 ; that is, x + y = 3. −3 −3 −2 −1 0 x 1 2 FIGURE 3.9.5 The curve sin(x + 2y) = 2x cos y and its tangent at the origin. ◗ 3 EXAMPLE 4 Figure 3.9.5 shows a computer plot of the graph of the equation sin(x + 2y) = 2x cos y. (4) Write the equation of the line tangent to this curve at the origin (0, 0). Solution When we differentiate each side in (4) with respect to the independent variable x, regarding y as a function of x, we get dy dy = 2 cos y − (2x sin y) . (5) [cos(x + 2y)] · 1 + 2 dx dx 196 Implicit Differentiation and Related Rates SECTION 3.9 197 We could collect coefficients and solve for the derivative dy/d x. But because we need only the slope y (0) at the origin, let us instead substitute x = y = 0 in Eq. (5). Noting that cos(0) = 1 and sin(0) = 0, we get the equation 1 + 2y (0) = 2, from which we see that y (0) = 12 . The resulting tangent line y = 12 x plotted in ◗ Fig. 3.9.5 “looks right,” and thus corroborates the results of our calculations. Related Rates A related-rates problem involves two or more quantities that vary with time and an equation that expresses some relationship between these quantities. Typically, the values of these quantities at some instant are given, together with all their time rates of change but one. The problem is usually to find the time rate of change that is not given, at some instant specified in the problem. Implicit differentiation, with respect to time t, of the equation that relates the given quantities will produce an equation that relates the rates of change of the given quantities. This is the key to solving a related-rates problem. EXAMPLE 5 Suppose that x(t) and y(t) are the x- and y-coordinates at time t of a point moving around the circle with equation x 2 + y 2 = 25. (6) Let us use the chain rule to differentiate both sides of this equation with respect to time t. This produces the equation 2x dy dx + 2y = 0. dt dt (7) If the values of x, y, and d x/dt are known at a certain instant t, then Eq. (7) can be solved for the value of dy/dt. It is not necessary to know x and y as functions of t. Indeed, it is common for a related-rates problem to contain insufficient information to express x and y as functions of t. For instance, suppose that we are given x = 3, y = 4, and d x/dt = 12 at a certain instant. Substituting these values into Eq. (7) yields 2 · 3 · 12 + 2 · 4 · dy = 0, dt so we find that dy/dt = −9 at the same instant. ◗ EXAMPLE 6 A rocket that is launched vertically is tracked by a radar station located on the ground 3 mi from the launch site. What is the vertical speed of the rocket at the instant that its distance from the radar station is 5 mi and this distance is increasing at the rate of 5000 mi/h? Solution Figure 3.9.6 illustrates this situation. We denote the altitude of the rocket (in miles) by y and its distance from the radar station by z. We are given dz = 5000 dt when z = 5. We want to find dy/dt (in miles per hour) at this instant. We apply the Pythagorean theorem to the right triangle in the figure and obtain y2 + 9 = z2 as a relation between y and z. From this we see that y = 4 when z = 5. Implicit differentiation then gives dz dy = 2z . 2y dt dt 197 198 CHAPTER 3 The Derivative dy dt z y 3 FIGURE 3.9.6 The rocket of Example 6. We substitute the data y = 4, z = 5, and dz/dt = 5000. Thus we find that dy = 6250 dt at the instant in question. (mi/h) ◗ Example 6 illustrates the following steps in the solution of a typical related-rates problem of the sort that involves a geometric situation: 1. Draw a diagram and label as variables the various changing quantities involved in the problem. 2. Record the values of the variables and their rates of change, as given in the problem. 3. Use the diagram to determine an equation that relates the important variables in the problem. 4. Differentiate this equation implicitly with respect to time t. 5. Substitute the given numerical data in the resulting equation, and then solve for the unknown. WARNING The most common error to be avoided is the premature substitution of the given data before differentiating implicitly. If we had substituted z = 5 to begin with in Example 5, our equation would have been y 2 + 9 = 25, and implicit differentiation would have given the absurd result dy/dt = 0. We use similar triangles (rather than the Pythagorean theorem ) in Example 7 to discover the needed relation between the variables. EXAMPLE 7 A man 6 ft tall walks with a speed of 8 ft/s away from a street light that is atop an 18-ft pole. How fast is the tip of his shadow moving along the ground when he is 100 ft from the light pole? Solution Let x be the man’s distance measured from the pole and z the distance from the tip of his shadow to the base of the pole (Fig. 3.9.7). Although x and z are positivevalued functions of time t, we do not attempt to find explicit formulas for either. We are given d x/dt = 8 (in feet per second); we want to find dz/dt when x = 100 (ft). We equate ratios of corresponding sides of the two similar triangles of Fig. 3.9.7 and find that z−x z = . 18 6 198 Implicit Differentiation and Related Rates SECTION 3.9 199 18 ft 6 ft Shadow z−x x z FIGURE 3.9.7 The moving shadow of Example 7. It follows that 2z = 3x, and implicit differentiation gives dz dx =3 . dt dt We substitute d x/dt = 8 and find that 2 dz 3 dx 3 = · = · 8 = 12. dt 2 dt 2 So the tip of the man’s shadow is moving at 12 ft/s. ◗ Example 7 is somewhat unusual in that the answer is independent of the man’s distance from the light pole—the given value x = 100 is superfluous because the tip of the man’s shadow is moving at constant speed. Example 8 is a related-rates problem with two relationships between the variables, which is not quite so unusual. EXAMPLE 8 Two radar stations at A and B, with B 6 km east of A, are tracking a ship. At a certain instant, the ship is 5 km from A, and this distance is increasing at the rate of 28 km/h. At the same instant, the ship is also 5 km from B, but this distance is increasing at only 4 km/h. Where is the ship, how fast is it moving, and in what direction is it moving? Solution With the distances indicated in Fig. 3.9.8, we find—again with the aid of the Pythagorean theorem—that Ship x 2 + y2 = u2 u Radar station A y x 6−x FIGURE 3.9.8 Radar stations tracking a ship (Example 8). Radar station B and (6 − x)2 + y 2 = v 2 . (8) We are given the following data: u = v = 5, du/dt = 28, and dv/dt = 4 at the instant in question. Because the ship is equally distant from A and B, it is clear that x = 3. Thus y = 4. Hence the ship is 3 km east and 4 km north of A. We differentiate implicitly the two equations in (8), and we obtain 2x and dy du dx + 2y = 2u dt dt dt dy dv dx + 2y = 2v . dt dt dt When we substitute the numerical data given and data deduced, we find that −2(6 − x) dx dy dx dy +4 = 140 and −3 +4 = 20. dt dt dt dt These equations are easy to solve: d x/dt = dy/dt = 20. Therefore, the ship is sailing northeast at a speed of √ 202 + 202 = 20 2 (km/h) 3 —if the figure is correct! A mirror along the line AB will √ reflect another ship, 3 km east and 4 km south of A, sailing southeast at a speed of 20 2 km/h. 199 200 CHAPTER 3 The Derivative The lesson? Figures are important, helpful, often essential—but potentially misleading. Avoid taking anything for granted when you draw a figure. In this example there would be no real problem, for each radar station could determine whether the ◗ ship was generally to the north or to the south. 3.9 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. x dy = . dx y Pierre de Fermat challenged René Descartes to find the line tangent to the graph of the folium x 2 + y 3 = 3x y at an arbitrary point. In a right triangle with short sides a and b and hypotenuse c, (a + b)2 = c2 . If two triangles have corresponding sides parallel, then the triangles are similar. In a related rates problem, one uses the fact that changes in related quantities are themselves related. Only one function is implicitly defined by the equation x 2 + y 2 = 100. Suppose that x = x(t) and y = y(t) are two functions of t such that x 2 + y 2 = 25 (for all t). If x = 3, y = 4, and x (t) = 12, then y (t) = −9. If x 3 + y 3 = 3x y, then Dx (x 3 ) = 3x 2 and D y (y 3 ) = 3y 2 . dy If x 3 + y 3 = 3x y, then Dx (3x y) = 3x + 3y. dx Folium is the Latin word for leaf. 1. If x 2 + y 2 = 100 then 2. 3. 4. 5. 6. 7. 8. 9. 10. 3.9 CONCEPTS: QUESTIONS AND DISCUSSION √ 1. Figure √ 3.9.1 shows the graphs of the two functions f2(x) = x and g(x) = − x that are defined implicitly by the equation x − y = 0. Both f and g are continuous for x 0. Can you think of a discontinuous function y = h(x) that satisfies the same equation? 2. How many different continuous functions of x (with the same domain of definition) are implicitly defined by a given quadratic equation in x and y? A given cubic equation? A given quartic (fourth-degree) equation? How many different discontinuous functions? 3. How many different continuous functions of x are defined by the following equations? (b) x 3 + y 3 = 1 (c) x 4 + y 4 = 1 (a) x 2 + y 2 + 1 = 0 4. How many different continuous functions of x are defined by the transcendental equation sin y = x? 3.9 PROBLEMS In Problems 1 through 4, first find the derivative dy/d x by implicit differentiation. Then solve the original equation for y explicitly in terms of x and differentiate to find dy/d x. Finally verify that your two results are the same by substituting the explicit expression for y(x) in the implicit form of the derivative. 1. x 2 − y 2 = 1 2. x y = 1 3. 16x 2 + 25y 2 = 400 4. x 3 + y 3 = 1 In Problems 5 through 14, find dy/d x by implicit differentiation. √ √ 6. x 4 + x 2 y 2 + y 4 = 48 5. x + y = 1 200 7. x 2/3 + y 2/3 = 1 8. (x − 1)y 2 = x + 1 9. x 2 (x − y) = y 2 (x + y) 10. x 5 + y 5 = 5x 2 y 2 11. x sin y + y sin x = 1 12. cos(x + y) = sin x sin y 13. 2x + 3e = e 14. x y = e−x y y x+y In Problems 15 through 28, use implicit differentiation to find an equation of the line tangent to the given curve at the given point. 15. x 2 + y 2 = 25; 16. x y = −8; (3, −4) (4, −2) Implicit Differentiation and Related Rates SECTION 3.9 17. x 2 y = x + 2; 18. x 1/4 +y 1/4 (2, 1) = 4; 28. y 2 = x 2 (x + 7); 2 21. 12(x 2 + y 2 ) = 25x y; (3, −2) 23. 2e−x + e y = 3e x−y ; (0, 0) 2 25. x 2/3 +y 2/3 5 y = 5; −10 (8, 1) (Fig. 3.9.9) −10 10 (8, 1) 0 −5 0 x 5 10 29. The curve x 3 + y 3 = 9x y is similar in shape and appearance to the folium of Descartes in Fig. 3.9.4. Find (a) the equation of its tangent line at the point (2, 4) and (b) the equation of its tangent line with slope −1. 30. (a) Factor the left-hand side of the equation −10 −10 −5 2x 2 − 5x y + 2y 2 = 0 0 x 5 10 FIGURE 3.9.9 Problem 25. 26. x 2 − x y + y 2 = 19; (3, −2) (Fig. 3.9.10) to show that its graph consists of two straight lines through the origin. Hence the derivative y (x) has only two possible numerical values (the slopes of these two lines). (b) Calculate dy/d x by implicit differentiation of the equation in part (a). Verify that the expression you obtain yields the proper slope for each of the straight lines of part (a). 31. Find all points on the graph of x 2 + y 2 = 4x + 4y at which the tangent line is horizontal. 10 8 6 4 2 y 0 −2 −4 −6 −8 −10 −10 32. Find the first-quadrant points of the folium of Example 3 at which the tangent line is either horizontal (dy/d x = 0) or vertical (where d x/dy = 1/(dy/d x) = 0). 33. Figure 3.9.13 shows the graph of the equation x = ye y . Show first that explicit differentiation to find d x/dy and implicit differentiation to find dy/d x yield consistent results. Then find the equation of the line tangent to the graph at the point (a) (0, 0); (b) (e, 1). (3, −2) −5 0 x 10 5 2 1.5 1 0.5 0 y −0.5 −1 −1.5 −2 −2.5 −3 −1 FIGURE 3.9.10 Problem 26. 27. (x 2 + y 2 )2 = 50x y; (2, 4) (Fig. 3.9.11) 6 (2, 4) 4 2 y 0 (e, 1) x = yey 0 1 2 3 4 x FIGURE 3.9.13 The curve x = ye y and its tangent line at the origin. −2 −4 −6 −6 −5 FIGURE 3.9.12 Problem 28. 5 y 0 −5 (3, 2) 24. x y = 6e2x−3y ; (−3, 6) 10 (3, 4) 22. x + x y + y = 7; 2 (Fig. 3.9.12) (16, 16) 19. x y + x y = 2; (1, −2) 1 1 + = 1; (1, 1) 20. x +1 y+1 2 (−3, 6) 201 −4 −2 0 x 2 4 FIGURE 3.9.11 Problem 27. 6 34. (a) Find the points on the curve x = ye y of Fig. 3.9.13 where the tangent line is vertical (d x/dy = 0). (b) Is there a point on the curve where the tangent line is horizontal? (c) Show that x → 0 and dy/d x → −∞ as y → −∞. (d) Show that y → +∞ and dy/d x → 0 as x → +∞. 201 202 CHAPTER 3 The Derivative 35. The graph in Fig. 3.9.14 is a lemniscate with equation (x 2 + y 2 )2 = x 2 − y 2 . Find by implicit differentiation the four points on the lemniscate where the tangent line is horizontal. Then find the two points where the tangent line is vertical—that is, where d x/dy = 1/(dy/d x) = 0. tank is 5 ft and is decreasing at the rate of 3 ft/s, at what rate is the radius r of the top surface of the water decreasing? y (x 2 + y 2 )2 = x 2 − y 2 10 r x y FIGURE 3.9.17 The spherical tank of Problem 38. FIGURE 3.9.14 The lemniscate of Problem 35. 36. Water is being collected from a block of ice with a square base (Fig. 3.9.15). The water is produced because the ice is melting in such a way that each edge of the base of the block is decreasing at 2 in./h while the height of the block is decreasing at 3 in./h. What is the rate of flow of water into the collecting pan when the base has edge length 20 in. and the height of the block is 15 in.? Make the simplifying assumption that water and ice have the same density. y x x FIGURE 3.9.15 The ice block of Problem 36. 37. Sand is being emptied from a hopper at the rate of 10 ft3 /s. The sand forms a conical pile whose height is always twice its radius (Fig. 3.9.16). At what rate is the radius of the pile increasing when its height is 5 ft? 39. A circular oil slick of uniform thickness is caused by a spill of 1 m3 of oil. The thickness of the oil slick is decreasing at the rate of 0.1 cm/h. At what rate is the radius of the slick increasing when the radius is 8 m? 40. Suppose that an ostrich 5 ft tall is walking at a speed of 4 ft/s directly toward a street light 10 ft high. How fast is the tip of the ostrich’s shadow moving along the ground? At what rate is the ostrich’s shadow decreasing in length? 41. The width of a rectangle is half its length. At what rate is its area increasing if its width is 10 cm and is increasing at 0.5 cm/s? 42. At what rate is the area of an equilateral triangle increasing if its base is 10 cm long and is increasing at 0.5 cm/s? 43. A gas balloon is being filled at the rate of 100π cm3 of gas per second. At what rate is the radius of the balloon increasing when its radius is 10 cm? 44. The volume V (in cubic inches) and pressure p (in pounds per square inch) of a certain gas satisfy the equation pV = 1000. At what rate is the volume of the sample changing if the pressure is 100 lb/in.2 and is increasing at the rate of 2 lb/in.2 per second? 45. Figure 3.9.18 shows a kite in the air at an altitude of 400 ft. The kite is being blown horizontally at the rate of 10 ft/s away from the person holding the kite string at ground level. At what rate is the string being payed out when 500 ft of string is already out? (Assume that the string forms a straight line.) 10 ft/s h = 2r 400 ft r FIGURE 3.9.16 The conical sand pile of Problem 37 with volume V = 13 πr 2 h. 38. Suppose that water is being emptied from a spherical tank of radius 10 ft (Fig. 3.9.17). If the depth of the water in the 202 Ground FIGURE 3.9.18 The kite of Problem 45. 46. A weather balloon that is rising vertically is being observed from a point on the ground 300 ft from the spot directly beneath the balloon. At what rate is the balloon rising when Implicit Differentiation and Related Rates SECTION 3.9 the angle between the ground and the observer’s line of sight is 45◦ and is increasing at 1◦ per second? 47. An airplane flying horizontally at an altitude of 3 mi and at a speed of 480 mi/h passes directly above an observer on the ground. How fast is the distance from the observer to the airplane increasing 30 s later? 48. Figure 3.9.19 shows a spherical tank of radius a partly filled with water. The maximum depth of water in the tank is y. A formula for the volume V of water in the tank— a formula you can derive after you study Chapter 6—is V = 13 π y 2 (3a − y). Suppose that water is being drained from a spherical tank of radius 5 ft at the rate of 100 gal/min. Find the rate at which the depth y of water is decreasing when (a) y = 7 (ft); (b) y = 3 (ft). [Note: One gallon of water occupies a volume of approximately 0.1337 ft3 .] Water level a y FIGURE 3.9.19 The spherical water tank of Problem 48. 49. Repeat Problem 48, but use a tank that is hemispherical, flat side on top, with radius 10 ft. 50. A swimming pool is 50 ft long and 20 ft wide. Its depth varies uniformly from 2 ft at the shallow end to 12 ft at the deep end (Fig. 3.9.20). Suppose that the pool is being filled at the rate of 1000 gal/min. At what rate is the depth of water at the deep end increasing when the depth there is 6 ft? [Note: One gallon of water occupies a volume of approximately 0.1337 ft3 .] 50 ft 2 ft Water level 12 ft y FIGURE 3.9.20 Cross section of the swimming pool of Problem 50. 51. A ladder 41 ft long that was leaning against a vertical wall begins to slip. Its top slides down the wall while its bottom moves along the level ground at a constant speed of 4 ft/s. How fast is the top of the ladder moving when it is 9 ft above the ground? 52. The base of a rectangle is increasing at 4 cm/s while its height is decreasing at 3 cm/s. At what rate is its area changing when its base is 20 cm and its height is 12 cm? 53. The height of a cone is decreasing at 3 cm/s while its radius is increasing at 2 cm/s. When the radius is 4 cm and the height is 6 cm, is the volume of the cone increasing or decreasing? At what rate is the volume changing then? 54. A square is expanding. When each edge is 10 in., its area is increasing at 120 in.2 /s. At what rate is the length of each edge changing then? 203 55. A rocket that is launched vertically is tracked by a radar station located on the ground 4 mi from the launch site. What is the vertical speed of the rocket at the instant its distance from the radar station is 5 mi and this distance is increasing at the rate of 3600 mi/h? 56. Two straight roads intersect at right angles. At 10 A . M . a car passes through the intersection headed due east at 30 mi/h. At 11 A . M . a truck heading due north at 40 mi/h passes through the intersection. Assume that the two vehicles maintain the given speeds and directions. At what rate are they separating at 1 P. M .? 57. A 10-ft ladder is leaning against a wall. The bottom of the ladder begins to slide away from the wall at a speed of 1 mi/h. (a) Find the rate at which the top of the latter is moving when it is 4 ft from the ground. If the top of the ladder maintained contact with the wall, find the speed with which it would be moving when it is (b) 1 in. above the ground; (c) 1 mm above the ground. Do you believe your answers? The key to the apparent paradox is that when the top of the ladder is about 1.65 ft high, it disengages altogether from the wall and thereafter slides away from it. 58. Two ships are sailing toward a very small island. One ship, the Pinta, is east of the island and is sailing due west at 15 mi/h. The other ship, the Niña, is north of the island and is sailing due south at 20 mi/h. At a certain time the Pinta is 30 mi from the island and the Niña is 40 mi from the island. At what rate are the two ships drawing closer together at that time? 59. At time t = 0, a single-engine military jet is flying due east at 12 mi/min. At the same altitude and 208 mi directly ahead of the military jet, still at time t = 0, a commercial jet is flying due north at 8 mi/min. When are the two planes closest to each other? What is the minimum distance between them? 60. A ship with a long anchor chain is anchored in 11 fathoms of water. The anchor chain is being wound in at the rate of 10 fathoms/min, causing the ship to move toward the spot directly above the anchor resting on the seabed. The hawsehole—the point of contact between ship and chain—is located 1 fathom above the water line. At what speed is the ship moving when there are exactly 13 fathoms of chain still out? 61. A water tank is in the shape of a cone with vertical axis and vertex downward. The tank has radius 3 ft and is 5 ft high. At first the tank is full of water, but at time t = 0 (in seconds), a small hole at the vertex is opened and the water begins to drain. When the height of the water in the tank has dropped to 3 ft, the water is flowing out at 2 ft3 /s. At what rate, in feet per second, is the water level dropping then? 62. A spherical tank of radius 10 ft is being filled with water at the rate of 200 gal/min. How fast is the water level rising when the maximum depth of water in the tank is 5 ft? See Problem 48 for a useful formula and a helpful note. 63. A water bucket is shaped like the frustum of a cone with height 2 ft, base radius 6 in., and top radius 12 in. Water is leaking from the bucket at 10 in.3 /min. At what rate is the water level falling when the depth of water in the bucket is 203 204 CHAPTER 3 The Derivative shows that the distance from the radar station to the plane is 5 mi and is increasing at 7 mi/min. What is the speed of the plane then (in miles per hour)? [Suggestion: You may find the law of cosines useful—see Appendix C.] 1 ft? [Note: The volume V of a conical frustum with height h and base radii a and b is V = πh 2 (a + ab + b2 ). 3 Such a frustum is shown in Fig. 3.9.21.] b h a FIGURE 3.9.21 The volume of this conical frustum is given in Problem 63. 64. Suppose that the radar stations A and B of Example 8 are now 12.6 km apart. At a certain instant, a ship is 10.4 km from A and its distance from A is increasing at 19.2 km/h. At the same instant, its distance from B is 5 km and is decreasing at 0.6 km/h. Find the location, speed, and direction of motion of the ship. 65. An airplane climbing at an angle of 45◦ passes directly over a ground radar station at an altitude of 1 mi. A later reading 66. The water tank of Problem 62 is completely full when a plug at its bottom is removed. According to Torricelli’s law, the √ water drains in such a way that d V /dt = −k y, where V is the volume of water in the tank and k is a positive empirical constant. (a) Find dy/dt as a function of the depth y. (b) Find the depth of water when the water level is falling the least rapidly. (You will need to compute the derivative of dy/dt with respect to y.) 67. A person 6 ft tall walks at 5 ft/s along one edge of a road 30 ft wide. On the other edge of the road is a light atop a pole 18 ft high. How fast is the length of the person’s shadow (on the horizontal ground) increasing when the person is 40 ft from the point directly across the road from the pole? 68. A highway patrol officer’s radar unit is parked behind a billboard 200 ft from a long straight stretch of U.S. 17. Down the highway, 200 ft from the point on the highway closest to the officer, is an emergency call box. The officer points the radar gun at the call box. A minivan passes the call box and, at that moment, the radar unit indicates that the distance between the officer and the minivan is increasing at 45 mi/h—that is, 66 ft/s. The posted speed limit is 55 mi/h. Does the officer have any reason to apprehend the driver of the minivan? 3.9 INVESTIGATION: Constructing the Folium of Descartes Computer graphics often requires lots of mathematics, and much mathematics was used in constructing many of the figures in this book. To see one way to construct Fig. 3.9.4, use a computer algebra system to solve the equation x 3 + y 3 = 3x y for y in terms of x. Verify that the three expressions you get define three different functions f , g, and h whose graphs are the three branches of the curve that are colored differently in Fig. 3.9.22. Investigate the domains of definition and the graphs of these functions to verify that they fit together precisely as shown in the figure. y 2 y = g(x) −2 y = f(x) 1 −1 1 −1 2 x y = h(x) −2 FIGURE 3.9.22 The equation x 3 + y 3 = 3x y implicitly defines three functions f , g, and h. 3.10 SUCCESSIVE APPROXIMATIONS AND NEWTON'S METHOD The solution of equations has always been a central task of mathematics. More than two millennia ago, mathematicians of ancient Babylon discovered the method of “completing the square,” which leads to the quadratic formula for an exact solution of any 204 Successive Approximations and Newton's Method SECTION 3.10 205 second-degree equation ax 2 +bx +c = 0. Early in the sixteenth century, several Italian mathematicians (Cardan, del Ferro, Ferrari, and Tartaglia) discovered formulas for the exact solutions of third- and fourth-degree equations. (Because they are quite complicated, these formulas are seldom used today except in computer algebra systems.) And in 1824 a brilliant young Norwegian mathematician, Niels Henrik Abel∗ (1802–1829), published a proof that there is no general formula giving the solution of an arbitrary polynomial equation of degree 5 (or higher) in terms of algebraic combinations of its coefficients. Thus the exact solution (for all its roots) of an equation such as f (x) = x 5 − 3x 3 + x 2 − 23x + 19 = 0 (1) may be quite difficult or even—as a practical matter—impossible to find. In such a case it may be necessary to resort to approximate methods. For example, the graph of y = f (x) in Fig. 3.10.1 indicates that Eq. (1) has three real solutions (and hence two complex ones as well). The indicated small rectangle 0.5 x 1, −5 y 5 encloses one of these solutions. If we use this small rectangle as a new “viewing window” with a computer or graphics calculator, then we see that this solution is near 0.8 (Fig. 3.10.2). A few additional magnifications might yield greater accuracy, showing that the solution is approximately 0.801. 80 y = f(x) 4 40 y y = f (x) 2 y 0 0 −2 −40 −4 −80 −2 0 x FIGURE 3.10.1 The graph y = f (x) in Eq. (1). 0.6 2 0.8 1 x FIGURE 3.10.2 Magnification of Fig. 3.10.1 near a solution. Graphical methods are good for three- or four-place approximations. Here we shall discuss an analytical method developed by Isaac Newton that can rapidly provide much more accurate approximations. Iteration and the Babylonian Square Root Method What it means to solve even so simple an equation as x2 − 2 = 0 (2) √ √ is open to question. The positive exact solution is x = 2. But the number 2 is irrational and hence cannot be expressed as a terminating or repeating decimal. Thus if we mean by a solution an exact decimal value for x, then even Eq. (2) can be solved only approximately. The ancient Babylonians devised √ an effective way to generate a sequence of better and better approximations to A, the square root of a given positive number A. Here square root method: We begin with a first guess x0 for √ √the value √ is the Babylonian of A. For 2, we might guess x0 = 1.5. If x0 is too large—that is, if x0 > A —then √ A A < √ = A, x0 A ∗ For the complete story of Abel’s remarkable achievements in his brief lifetime, see Oystein Ore’s very readable biography Niels Henrik Abel (The University of Minnesota and Chelsea Publishing Company, 1974). 205 206 CHAPTER 3 The Derivative √ √ so A/x0 is too small an estimate of √A. Similarly, if x0 √ is too small (if x0 < A), then A/x0 is too large an estimate of A; that is, A/x0 > A. √ Thus in each case one of the two numbers x0 and A/x0 is an underestimate of A and the other√ is an overestimate. The Babylonian idea was that we should get a better estimate of A by averaging x0 and A/x0 . This yields a better approximation 1 A x1 = x0 + (3) 2 x0 √ to A. But why not repeat this process? We can average x1 and A/x1 to get a second approximation x2 , average x2 and A/x2 to get x3 , and so on. By repeating this process, we generate a sequence of numbers x1 , x2 , x3 , x4 , ... that √ we have every right to expect will consist of better and better approximations to A. Specifically, having calculated the nth approximation xn , we calculate the next one by means of the iterative formula xn+1 1 A = xn + . 2 xn (4) √ In other words, we plow each approximation to A back into the right-hand side in Eq. (4) to calculate the next approximation. This is an iterative process—the words iteration and iterative are derived from the Latin iterare, “to plow again.” Suppose we find that after sufficiently many steps in this iteration, xn+1 ≈ xn accurate to the number of decimal places we are retaining in our computations. Then Eq. (4) yields 1 1 A xn + = x2 + A , xn ≈ xn+1 = 2 xn 2xn n so 2xn2 ≈ xn2 + A, and hence xn2 ≈ A to some degree of accuracy. EXAMPLE 1 With A = 2 we begin with the crude first guess x0 = 1 to the value √ of A. Then successive applications of the formula in Eq. (4) yield 1 2 3 x1 = 1+ = = 1.5, 2 1 2 2 17 1 3 + = ≈ 1.416666667, x2 = 2 2 3/2 12 1 17 2 577 x3 = + = ≈ 1.414215686, 2 12 17/12 408 2 665857 1 577 + = ≈ 1.414213562, x4 = 2 408 577/408 470832 rounding results to nine decimal places. It happens that x4 gives nine places! √ 2 accurate to all ◗ The Babylonian iteration defined in Eq.√(4) is a method for generating a sequence of approximations to the positive root r = A of the particular equation x 2 − A = 0. We turn next to a method that gives such a sequence of approximations for more general equations. 206 Successive Approximations and Newton's Method SECTION 3.10 207 Newton's Method Newton’s method is an iterative method for generating a sequence x1 , x2 , x3 , . . . of approximations to a solution r of a given equation written in the general form f (x) = 0. (5) We hope that this sequence of approximations will “converge” to the root r in the sense of the following definition. DEFINITION Convergence of Approximations We say that the sequence of approximations x1 , x2 , x3 , . . . converges to the number r provided that we can make xn as close to r as we please merely by choosing n sufficiently large. More precisely, for any given > 0, there exists a positive integer N such that |xn − r | < for all n N . As a practical matter such convergence means, as illustrated in Example 1, that for any positive integer k, xn and r will agree to k or more decimal places once n becomes sufficiently large. The idea is that we begin with an initial guess x0 that roughly approximates a solution r of the equation f (x) = 0. This initial guess may, for example, be obtained by inspection of the graph of y = f (x), perhaps obtained from a computer or graphics calculator. We use x0 to calculate an approximation x1 , use x1 to calculate a better approximation x2 , use x2 to calculate a still better approximation x3 , and so on. Here is the general step in the process. Having reached the nth approximation xn , we use the tangent line at (xn , f (xn )) to construct the next approximation xn+1 to the solution r as follows: Begin at the point xn on the x-axis. Go vertically up (or down) to the point (xn , f (xn )) on the curve y = f (x). Then follow the tangent line L there to the point where L meets the x-axis (Fig. 3.10.3). That point will be xn+1 . y y = f(x) L (xn, f (xn)) r xn + 1 xn x FIGURE 3.10.3 Geometry of the formula of Newton’s method. Here is a formula for xn+1 . We obtain it by computing the slope of the line L in two ways: from the derivative and from the two-point definition of slope. Thus f (xn ) = f (xn ) − 0 , xn − xn+1 and we easily solve for xn+1 = xn − f (xn ) . f (xn ) (6) This equation is the iterative formula of Newton’s method, so called because in about 1669, Newton introduced an algebraic procedure (rather than the geometric construction illustrated in Fig. 3.10.3) that is equivalent to the iterative use of Eq. (6). Newton’s 207 208 CHAPTER 3 The Derivative first example was the cubic equation x 3 − 2x − 5 = 0, for which he found the root r ≈ 2.0946 (as we ask you to do in Problem 18). Suppose now that we want to apply Newton’s method to solve the equation f (x) = 0 (7) to an accuracy of k decimal places (k digits to the right of the decimal correct or correctly rounded). Remember that an equation must be written precisely in the form of Eq. (7) in order to use the formula in Eq. (6). If we reach the point in our iteration at which xn and xn+1 agree to k decimal places, it then follows that xn ≈ xn+1 = xn − f (xn ) ; f (xn ) 0≈− f (xn ) ; f (xn ) f (xn ) ≈ 0. Thus we have found an approximate root xn ≈ xn+1 of Eq. (7). In practice, then, we retain k decimal places in our computations and persist until xn ≈ xn+1 to this degree of accuracy. (We do not consider here the possibility of round-off error, an important topic in numerical analysis.) √ EXAMPLE 2 Use Newton’s method to find 2 accurate to nine decimal places. Solution More generally, consider the square root of the positive number A as the positive root of the equation f (x) = x 2 − A = 0. Because f (x) = 2x, Eq. (6) gives the iterative formula xn2 − A 1 A xn+1 = xn − . = xn + 2xn 2 xn x 11 − 2x 7− Thus we have derived the Babylonian iterative formula as a special case of Newton’s method. The use of Eq. (8) with A = 2 therefore yields exactly the values of x1 , x2 , x3 , and x4 that we computed in Example 1, and after performing another iteration we find that 1 2 x4 + ≈ 1.414213562, x5 = 2 x4 which agrees with x4 to nine decimal places. The very rapid convergence here is an important characteristic of Newton’s method. As a general rule (with some exceptions), ◗ each iteration doubles the number of decimal places of accuracy. 2x FIGURE 3.10.4 The tray of Example 3. EXAMPLE 3 Figure 3.10.4 shows an open-topped tray constructed by the method of Example 2 in Section 3.6. We begin with a 7-by-11-in. rectangle of sheet metal. We cut a square with edge length x from each of its four corners and then fold up the resulting flaps to obtain a rectangular tray with volume V (x) = x(7 − 2x)(11 − 2x) y = 4x3 − 36x2 + 77x − 40 = 4x 3 − 36x 2 + 77x, 40 y (8) 0 x 3.5. (9) In Section 3.6 we inquired about the maximum possible volume of such a tray. Here we want to find instead the value(s) of x that will yield a tray with volume 40 in.3 ; we will find x by solving the equation 0 V (x) = 4x 3 − 36x 2 + 77x = 40. −40 To solve this equation for x, first we write an equation of the form in Eq. (7): 0 4 8 x FIGURE 3.10.5 The graph of f (x) in Eq. 10 of Example 3. f (x) = 4x 3 − 36x 2 + 77x − 40 = 0. Figure 3.10.5 shows the graph of f . We see three solutions: a root r1 between 0 and 1, a root r2 slightly greater than 2, and a root r3 slightly larger than 6. Because f (x) = 12x 2 − 72x + 77, 208 (10) Successive Approximations and Newton's Method SECTION 3.10 209 Newton’s iterative formula in Eq. (6) takes the form f (xn ) f (xn ) 4x 3 − 36xn2 + 77xn − 40 . = xn − n 12xn2 − 72xn + 77 xn+1 = xn − (11) Beginning with the initial guess x0 = 1 (because it’s reasonably close to r1 ), Eq. (11) gives 4 · 13 − 36 · 12 + 77 · 1 − 40 ≈ 0.7059, 12 · 12 − 72 · 1 + 77 x2 ≈ 0.7736, x3 ≈ 0.7780, x4 ≈ 0.7780. x1 = 1 − Thus we obtain the root r1 ≈ 0.7780, retaining only four decimal places. If we had begun with a different initial guess, the sequence of Newton iterates might well have converged to a different root of the equation f (x) = 0. The approximate solution obtained therefore depends on the initial guess. For example, with x0 = 2 and, later, with x0 = 6, the iteration in Eq. (11) produces the two sequences x0 x1 x2 x3 x4 2 y=x y = (1/2) cos x y 0 −2 −2 0 x 2 FIGURE 3.10.6 Solving the equation x = 12 cos x (Example 4). =2 ≈ 2.1053 ≈ 2.0993 ≈ 2.0992 ≈ 2.0992 x0 x1 x2 x3 x4 =6 ≈ 6.1299 ≈ 6.1228 ≈ 6.1227 ≈ 6.1227 Thus the other two roots of Eq. (10) are r2 ≈ 2.0992 and r3 ≈ 6.1277. With x = r1 ≈ 0.7780, the tray in Fig. 3.10.4 has the approximate dimensions 9.4440 in. by 5.4440 in. by 0.7780 in. With x = r2 ≈ 2.0992, its approximate dimensions are 6.8016 in. by 2.8016 in. by 2.0992 in. But the third root r3 ≈ 6.1227 would not lead to a tray that is physically possible. (Why not?) Thus the two values of x that ◗ yield trays with volume 40 in.3 are x ≈ 0.7780 and x ≈ 2.0992. EXAMPLE 4 Figure 3.10.6 indicates that the equation x= 1 2 cos x (12) has a solution r near 0.5. To apply Newton’s method to approximate r , we rewrite Eq. (12) in the form f (x) = 2x − cos x = 0. Because f (x) = 2 + sin x, the iterative formula of Newton’s method is 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 −5 y = ln x ? xn+1 = xn − 2xn − cos xn . 2 + sin xn Beginning with x0 = 0.5 and retaining five decimal places, this formula yields x1 ≈ 0.45063, x2 ≈ 0.45018, x3 ≈ 0.45018. ◗ Thus the root is 0.45018 to five decimal places. y = 3 sin x 0 5 10 15 20 25 30 x FIGURE 3.10.7 The graphs y = 3 sin x and y = ln x. EXAMPLE 5 Figure 3.10.7 indicates that the equation 3 sin x = ln x has either five or six positive solutions. To better approximate the smallest solution r ≈ 3, we apply Newton’s method with f (x) = 3 sin x − ln x, f (x) = 3 cos x − 1 . x 209 210 CHAPTER 3 The Derivative Then the iterative formula of Newton’s method is 3 sin xn − ln xn . xn+1 = xn − 3 cos xn − (1/xn ) When we begin with x0 = 3 and retain five decimal places, this formula gives x1 ≈ 2.79558, x2 ≈ 2.79225, x3 ≈ 2.79225. Thus r ≈ 2.79225 to five decimal places. In Problem 42 we ask you to find the remaining solutions indicated in Fig. 3.10.7. ◗ EXAMPLE 6 Newton’s method is one for which “the proof is in the pudding.” If it works, it’s obvious that it does, and everything’s fine. When Newton’s method fails, it may do so spectacularly. For example, suppose that we want to solve the equation x 1/3 = 0. Here r = 0 is the only solution. The iterative formula in Eq. (6) becomes xn+1 = xn − (xn )1/3 = xn − 3xn = −2xn . 1 (x )−2/3 3 n If we begin with x0 = 1, Newton’s method yields x1 = −2, x2 = +4, x3 = −8, and ◗ so on. Figure 3.10.8 indicates why our “approximations” are not converging. y y = x 1/3 −2 1 4 x FIGURE 3.10.8 A failure of Newton’s method. When Newton’s method fails, a graph will typically indicate the reason why. Then the use of an alternative method such as repeated tabulation or successive magnification is appropriate. Newton's Method with Calculators and Computers With calculators and computers that permit user-defined functions, Newton’s method is very easy to set up and apply repeatedly. It is helpful to interpret Newton’s iteration xn+1 = xn − f (xn ) f (xn ) as follows. Having first defined the functions f and f , we then define the “iteration function” f (x) g(x) = x − . f (x) Newton’s method is then equivalent to the following procedure. Begin with an initial estimate x0 of the solution of the equation f (x) = 0. Calculate successive approximations x1 , x2 , x3 , . . . to the exact solution by means of the iteration xn+1 = g(xn ). That is, apply the function g to each approximation to get the next. 210 Successive Approximations and Newton's Method SECTION 3.10 211 Figure 3.10.9 shows a TI graphics calculator prepared to solve the equation f (x) = x 3 − 3x 2 + 1 = 0. Then we need only store the initial guess, 0.5 → x, and next enter repeatedly the command y3 → x, as indicated in Fig. 3.10.10. TEXAS INSTRUMENTS TI-85 t t TEXAS INSTRUMENTS TI-85 FIGURE 3.10.9 Preparing to solve the equation x 3 − 3x 2 + 1 = 0. 48SX FIGURE 3.10.10 Solving the equation x 3 − 3x 2 + 1 = 0. SCIENTIFIC EXPANDABLE FIGURE 3.10.11 Preparing to solve the equation x 3 − 3x 2 + 1 = 0. Figure 3.10.11 shows an HP calculator prepared to carry out the same iteration. The functions F(X), D(X) (for f (x)), and G(X) are each defined by pressing the DEFINE key. Then it is necessary only to ENTER the initial guess x 0 and press the G key repeatedly to generate the desired successive appropriations. With Maple or Mathematica you can define the functions f and g and then repeatedly enter the command x = g(x), as shown in Fig. 3.10.12. Mathematica Command Maple Command f[x ] := x∧ 3 − 3x∧ 2 + 1 g[x ] := x − f[x]/f [x] x = 0.5 x = g[x] x = g[x] x = g[x] x = g[x] f : = x −> x∧ 3 − 3 ∗ x ∧ 2 + 1; g : = x −> x − f(x)/D(f)(x); x : = 0.5; x : = g(x); x : = g(x); x : = g(x); x : = g(x); Result 0.500000 0.666667 0.652778 0.652704 0.652704 FIGURE 3.10.12 Mathematica and Maple implementations of Newton’s method. Newton's Method and Computer Graphics Newton’s method and similar iterative techniques are often used to generate vividly colored “fractal patterns,” in which the same or similar structures are replicated on smaller and smaller scales at successively higher levels of magnification. To describe one way this can be done, we replace the real numbers in our Newton’s method computations with complex numbers. We illustrate this idea with the cubic equation f (x) = x 3 − 3x 2 + 1 = 0. (13) In the Investigation we ask you to approximate the three solutions r1 ≈ −0.53, r2 ≈ 0.65, r3 ≈ 2.88 of this equation. √ First,2 recall that a complex number is a number of the form a + bi, where i = −1, so i = −1. The real numbers a and b are called the real part and the imaginary part, respectively, of a + bi. You add, multiply, and divide complex numbers as if they were binomials, with real and imaginary parts “collected” as in the computations (3 + 4i) + (5 − 7i) = (3 + 5) + (4 − 7)i = 8 − 3i, (2 + 5i)(3 − 4i) = 2(3 − 4i) + 5i(3 − 4i) = 6 − 8i + 15i − 20i 2 = 26 + 7i, 211 212 CHAPTER 3 The Derivative and 26 + 7i 2 + 5i 3 − 4i 26 + 7i 2 + 5i = = · = = 1.04 + (0.28)i. 2 3 + 4i 3 + 4i 3 − 4i 9 − 16i 25 The use of the conjugate 3 − 4i of the denominator 3 + 4i in the last computation is a very common technique for writing a complex fraction in the standard form a + bi. (The conjugate of x + yi is x − yi; it follows that the conjugate of x − yi is x + yi.) Now let us substitute the complex number z = x + i y into the cubic polynomial f (z) = z 3 − 3z 2 + 1 of Eq. (13) and into its derivative f (z) = 3z 2 − 6z. We find that f (z) = (x + i y)3 − 3(x + i y)2 + 1 = (x 3 − 3x y 2 − 3x 2 + 3y 2 + 1) + (3x 2 y − y 3 − 6x y)i (14) and f (z) = 3(x + i y)2 − 6(x + i y) = (3x 2 − 3y 2 − 6x) + (6x y − 6y)i. (15) Consequently, there is nothing to prevent us from applying Newton’s method to Eq. (13) with complex numbers. Beginning with a complex initial guess z 0 = x0 + i y0 , we can substitute Eqs. (14) and (15) into Newton’s iterative formula z n+1 = z n − f (z n ) f (z n ) (16) to generate the complex sequence {z n }, which may yet converge to a (real) solution of Eq. (13). With this preparation, we can now explain how Fig. 3.10.13 was generated. A computer was programmed to carry out Newton’s iteration repeatedly, beginning with many thousands of initial guesses z 0 = x0 + i y0 that “fill” the rectangle −2 x 4, −2.25 y 2.25 in the complex plane. The initial point z 0 = x0 + i y0 was then color-coded according to the root (if any) to which the corresponding sequence {z n } converged: Color z 0 green if {z n } converges to the root r1 ≈ −0.53; Color z 0 red if {z n } converges to the root r2 ≈ 0.65; Color z 0 yellow if {z n } converges to the root r3 ≈ 2.88. FIGURE 3.10.13 −2 x 4, −2.25 y 2.25. Thus we use different colors to distinguish different “Newton basins of attraction” for the equation we are investigating. It is not surprising that a red region containing the root r2 appears in the middle of Fig. 3.10.13, separating a green region to the left that contains r1 and a yellow region to the right that contains r3 . But why would yellow lobes protrude from the green region into the red region and green lobes protrude from the yellow region into the red one? To see what’s happening near these lobes, we generated some blowups. 212 Successive Approximations and Newton's Method SECTION 3.10 213 Figure 3.10.14 shows a blowup of the rectangle 1.6 x 2.4, −0.3 y 0.3 containing the green lobe that’s visible in Fig. 3.10.13. Figure 3.10.15 (1.64 x 1.68, −0.015 y 0.015) and Fig. 3.10.16 (1.648 x 1.650, −0.00075 y 0.00075) are further magnifications. The rectangle shown in Fig. 3.10.16 corresponds to less than one millionth of a square inch of Fig. 3.10.13. FIGURE 3.10.14 1.6 x 2.4, −0.3 y 0.3. FIGURE 3.10.15 1.64 x 1.68, −0.015 y 0.015. FIGURE 3.10.16 1.648 x 1.650, −0.00075 y 0.00075. At every level of magnification, each green lobe has smaller yellow lobes protruding into the surrounding red region, and each of these yellow lobes has still smaller green lobes protruding from it, and so on ad infinitum (just like the proverbial little fleas that are bitten by still smaller fleas, and so on ad infinitum). Figure 3.10.17 shows the Newton basins picture for the twelfth-degree polynomial equation f (x) = x 12 − 14x 10 + 183x 8 − 612x 6 − 2209x 4 − 35374x 2 + 38025 = 0, (17) which has as its solution the twelve complex numbers 1, 3, 1 ± 2i, 3 ± 2i, − 1, − 3, −1 ± 2i, −3 ± 2i. Twelve different colors are used to distinguish the Newton basins of these twelve solutions of Eq. (17). Where the fractal common boundary appears to separate basins of different colors, it is studded with “flowers” like the one at the center of Fig. 3.10.17, which is magnified in Fig. 3.10.18. Each of these flowers has ten “leaves” (in the remaining ten colors). Each of these leaves has “buds” like the one shown in Fig. 3.10.19. Each of these buds is encircled with flowers that have leaves that have buds that are encircled with flowers—and so on ad infinitum. FIGURE 3.10.17 Newton basis for the twelfth-degree polynomial. FIGURE 3.10.18 The flower at the center of Fig. 3.10.17. FIGURE 3.10.19 A bud on a petal of the flower in Fig. 3.10.18. 3.10 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Neils Henrik Abel lived early in the nineteenth century. 213 214 CHAPTER 3 The Derivative √ 2. In the Babylonian square root method for approximating 2, beginning with the first approximation x0 = 1 leads to the second approximation x1 = 1.5. 3. To use Newton’s method to solve an equation in (the single variable) x, first write the equation in the form f (x) = 0. 4. The formula used in Newton’s method can be derived with the aid of Fig. 3.10.3. 5. The iterative formula of Newton’s method is f (xn ) xn+1 = xn + for n 0. f (xn ) √ 6. One application of Newton’s method to approximate 2 using the initial approx. imation x0 = 32 yields x1 = 17 12 7. Suppose that you use Newton’s method to approximate a solution of the equation f (x) = 0 and you find that the approximations xn and xn+1 agree to 100 decimal places. Then you can be quite sure that either is an excellent approximation to a solution of f (x) = 0. 8. The smallest positive solution of 4x 3 − 36x 2 + 77x − 40 = 0 is approximately 0.7780. 9. The largest positive solution of 4x 3 − 36x 2 + 77x − 40 = 0 is approximately 6.1227. 10. The only positive solution of 2x = cos x is approximately 0.45018. 3.10 CONCEPTS: QUESTIONS AND DISCUSSION 1. Example 1 in this section illustrates the use of Babylonian iteration to approximate the square root of a positive number A, beginning with a positive initial guess x0 . How is the number of iterations required for six-place accuracy affected by choosing x0 very close to zero, or very large? Does it appear that the number of decimal places of accuracy is roughly doubled with each iteration? What happens if a negative initial guess is used? What happens if A itself is negative? 2. The general rule—that each iteration of Newton’s method typically doubles the number of decimal places of accuracy—does not hold when the method is used to approximate a solution r of f (x) = 0 if r is also a critical point of f . Investigate the “rate of convergence” to the root r if: (a) f (x) = (x − 2)2 , so that r = 2 is a double root; (b) f (x) = (x − 1)2/3 , so that the graph has a cusp and a vertical tangent at r = 1. 3. Consider the exotic function f (x) = x 2 sin(1/x) [with f (0) = 0] of Problem 88 in Section 3.7. Investigate what happens when you use a computer algebra system to attempt to approximate the root r = 0 by iterating g(x) = x − f (x)/ f (x). Try a variety of different nonzero initial guesses and explain the results. 3.10 PROBLEMS In Problems 1 through 20, use Newton’s method to find the solution of the given equation f (x) = 0 in the indicated interval [a, b] accurate to four decimal places. You may choose the initial guess either on the basis of a calculator graph or by interpolation between the values f (a) and f (b). [−1, 0] 8. x + 3x + 2x = 10; 3 [1, 2] 2 9. x − cos x = 0; [0, 2] 10. x 2 − sin x = 0; [0.5, 1.0] [1, 2] 1. x − 5 = 0; [2, 3] (to find the positive square root of 5) 11. 4x − sin x = 4; 2. x − 2 = 0; [1, 2] (to find the cube root of 2) 12. 5x + cos x = 5; [0, 1] [2, 3] 2 3 3. x − 100 = 0; [2, 3] (to find the fifth root of 100) 13. x 5 + x 4 = 100; 4. x 3/2 − 10 = 0; [4, 5] (to find 102/3 ) 14. x 5 + 2x 4 + 4x = 5; 5 [0, 1] 5. x + 3x − 1 = 0; [0, 1] 15. x + tan x = 0; [2, 3] 6. x 3 + 4x − 1 = 0; [0, 1] 16. x + tan x = 0; [11, 12] 2 214 7. x 6 + 7x 2 − 4 = 0; Successive Approximations and Newton's Method SECTION 3.10 215 17. 18. 19. 20. 21. x − e−x = 0; [0, 1] x 3 − 2x − 5 = 0; [2, 3] (Newton’s own example) e x + x − 2 = 0; [0, 1] e−x − ln x = 0; [1, 2] (a) Show that Newton’s method applied to the equation x 3 − a = 0 yields the formula 1 a 2xn + 2 xn+1 = 3 xn for approximating the cube root of a. (b) Use this formula √ 3 to find 2 accurate to five decimal places. 22. (a) Show that Newton’s method yields the formula xn+1 = 1 a (k − 1)xn + k (xn )k−1 for approximating the kth√ root of the positive number a. (b) Use this formula to find 10 100 accurate to five decimal places. 23. Equation (12) has the special form x = G(x), where G(x) = 1 cos x. For an equation of this form, the iterative formula 2 xn+1 = G(xn ) produces a sequence of approximations that sometimes converges to a root. In the case of Eq. (12), this repeated substitution formula is simply xn+1 = 12 cos xn . Begin with x0 = 0.5 as in Example 4 and retain five decimal places in your computation of the solution of Eq. (12). [Check: You should find that x8 ≈ 0.45018.] 24. The equation x 4 = x + 1 has a solution between x = 1 and x = 2. Use the initial guess x0 = 1.5 and the method of repeated substitution (see Problem 23) to discover that one of the solutions of this equation is approximately 1.220744. Iterate using the formula and thus provides a method for approximating the reciprocal 1/a without performing any divisions. Such a method is useful because, in most high-speed computers, the operation of division is more time consuming than even several additions and multiplications. 27. Prove that the equation x 5 + x = 1 has exactly one real solution. Then use Newton’s method to find it with four places correct to the right of the decimal point. In Problems 28 through 30, use Newton’s method to find all real roots of the given equation with four digits correct to the right of the decimal point. [Suggestion: In order to determine the number of roots and their approximate locations, graph the left- and right-hand sides of each equation and observe where the graphs cross.] 28. x 2 = cos x 29. x = 2 sin x 30. cos x = − 15 x (There are exactly three solutions, as indicated in Fig. 3.10.20.) y y = cos x x y=−x 5 FIGURE 3.10.20 Solving the equation in Problem 30. Then compare the result with what happens when you iterate using the formula 31. Prove that the equation x 7 − 3x 3 + 1 = 0 has at least one solution. Then use Newton’s method to find one solution to three-place accuracy. √ 32. Use Newton’s method to approximate 3 5 to four-place accuracy. xn+1 = (xn )4 − 1. 33. Use Newton’s method to find the value of x for which x 3 = cos x. xn+1 = (xn + 1)1/4 . 25. The equation x 3 − 3x 2 + 1 = 0 has a solution between x = 0 and x = 1. To apply the method of repeated substitution (see Problem 23) to this equation, you may write it either in the form x =3− 1 x2 34. Use Newton’s method to find the smallest positive value of x for which x = tan x. 35. In Problem 49 of Section 3.6, we dealt with the problem of minimizing the cost of building a road to two points on opposite sides of a geologic fault. This problem led to the equation f (x) = 3x 4 − 24x 3 + 51x 2 − 32x + 64 = 0. or in the form x = (3x 2 − 1)1/3 . If you begin with x0 = 0.5 in the hope of finding the nearby solution (approximately 0.6527) of the original equation by using each of the preceding iterative formulas, you will observe some of the drawbacks of the method. Describe what goes wrong. 26. Show that Newton’s method applied to the equation 1 −a =0 x yields the iterative formula xn+1 = 2xn − a(xn )2 Use Newton’s method to find, to four-place accuracy, the root of this equation that lies in the interval [3, 4]. 36. The moon of Planet Gzyx has an elliptical orbit with eccentricity 0.5, and its period of revolution around the planet is 100 days. If the moon is at the position (a, 0) when t = 0, then (Fig. 3.10.21) the central angle after t days is given by Kepler’s equation 2πt 1 = θ − sin θ. 100 2 Use Newton’s method to solve for θ when t = 17 (days). Take θ0 = 1.5 (rad) and calculate the first two approximations θ1 and θ2 . Express θ2 in degrees as well. 215 216 CHAPTER 3 The Derivative )0, a 2 3 ) Finally, use Newton’s method to find first the possible values of t, and then those of w, accurate to four decimal places. Moon (a, 0) θ 20 15 Planet, a at ( , 0) 2 y x 5 u FIGURE 3.10.21 The elliptical orbit of Problem 36. 37. A great problem of Archimedes was that of using a plane to cut a sphere into two segments with volumes in a given (preassigned) ratio. Archimedes showed that the volume of a segment of height h of a sphere of radius a is V = 1 π h 2 (3a − h). If a plane at distance x from the cen3 ter of a sphere of radius 1 cuts the sphere into two segments, one with twice the volume of the other, show that 3x 3 − 9x + 2 = 0. Then use Newton’s method to find x accurate to four decimal places. 38. The equation f (x) = x 3 − 4x + 1 = 0 has three distinct real roots. Approximate their locations by evaluating f at x = −3, −2, −1, 0, 1, 2, and 3. Then use Newton’s method to approximate each of the three roots to four-place accuracy. 39. The equation x + tan x = 0 is important in a variety of applications—for example, in the study of the diffusion of heat. It has a sequence α1 , α2 , α3 , . . . of positive roots, with the nth root slightly larger than (n − 0.5)π . Use Newton’s method to compute α1 and α2 to three-place accuracy. 40. Investigate the cubic equation 4x 3 − 42x 2 − 19x − 28 = 0. Perhaps you can see graphically that it has only a single real solution. Find it (accurate to four decimal places). First try the initial guess x0 = 0; be prepared for at least 25 iterations. Then try initial guesses x0 = 10 and x0 = 100. 41. A 15-ft ladder and a 20-ft ladder lean in opposite directions against the vertical walls of a hall (Fig. 3.10.22). The ladders cross at a height of 5 ft. You must find the width w of the hall. First, let x and y denote the heights of the tops of the ladders on the walls and u and v the lengths shown in the figure, so that w = u + v. Use similar triangles to show that u v x =5 1+ , y =5 1+ . v u Then apply the Pythagorean theorem to show that t = u/v satisfies the equation t 4 + 2t 3 + 7t 2 − 2t − 1 = 0. 216 FIGURE 3.10.22 The crossing ladders of Problem 41. 42. Use Newton’s method to find the remaining positive solutions of the equation 3 sin x = ln x of Example 5 (Fig. 3.10.7). Do whatever is necessary to determine whether there is or is not a solution near x = 20. 43. The spherical asteroid problem in Problem 49 in Section 1.4 leads to the equation (100 + θ) cos θ = 100, where R = 1000/θ is the radius of the asteroid, and it is clear from the context that 0 < θ < π/2. Use Newton’s method to solve this problem. 44. This is a famous “railroad track problem.” Consider a 1-mile railroad track that was constructed without leaving the usual expansion spaces between consecutive rails. Thus each rail of the track is, in effect, a single steel rail one mile long. Suppose that an increase in the temperature by 20◦ C increases—by thermal expansion of the steel—the length of this rail by one foot. Also assume that the ends of the track are fixed, so the rail “bows up” in the shape of a circular arc with central angle 2θ and radius R (Fig. 3.10.23). Find the resulting height x (at its midpoint) of the bowed rail above the ground. ed track Curv Straight track x R θ θ FIGURE 3.10.23 The bowed railroad track of Problem 44. Successive Approximations and Newton's Method SECTION 3.10 217 3.10 INVESTIGATION: How Deep Does a Floating Ball Sink? Figure 3.10.24 shows a large cork ball of radius a = 1 floating in water of density 1. If the ball’s density ρ is one-fourth that of water, ρ = 14 , then Archimedes’ law of buoyancy implies that the ball floats in such a way that one-fourth of its total volume is submerged. Because the volume of the ball is 4π/3, it follows that the volume of the part of the ball beneath the waterline is given by V =ρ· 1 4π π 4π = · = . 3 4 3 3 (1) 1 1–x r Waterline x FIGURE 3.10.24 The floating cork ball. The shape of the submerged part of the ball is that of a spherical segment with a circular flat top. The volume of a spherical segment of top radius r and depth h = x (as in Fig. 3.10.24) is given by the formula πx V = (3r 2 + x 2 ). (2) 6 This formula is also due to Archimedes and holds for any depth x, whether the spherical segment is smaller or larger than a hemisphere. For instance, note that with r = 0 and x = 2a it gives V = 43 πa 3 , the volume of an entire sphere of radius a. For a preliminary investigation, proceed as follows to find the depth x to which the ball sinks in the water. Equate the two expressions for V in Eqs. (1) and (2), then use the right triangle in Fig. 3.10.24 to eliminate r . You should find that x must be a solution of the cubic equation f (x) = x 3 − 3x 2 + 1 = 0. As the graph y = f (x) in Fig. 3.10.25 indicates, this equation has three real solutions—one in (−1, 0), one in (0, 1), and one in (2, 3). The solution between 0 and 1 gives the actual depth x to which the ball sinks (why?). You can find x using Newton’s method. 4 2 (3) y = x3 − 3x2 + 1 Your Investigation For your very own floating ball to investigate, let its density ρ in Eq. (1) be given by y 0 −2 10 + k 20 where k denotes the last nonzero digit in the sum of the final four digits of your student I . D . number. Your objective is to find the depth to which this ball sinks in the water. Begin by deriving the cubic equation that you need to solve, explaining each step carefully. Then find all of its solutions accurate to at least four decimal places. Include in your report a sketch of a spherical ball with the waterline located accurately (to scale) in the position corresponding to your result for the desired depth. ρ= −4 −4 −2 0 x 2 FIGURE 3.10.25 Graph for the cork-ball equation. 4 217 218 CHAPTER 3 The Derivative CHAPTER 3: REVIEW Understanding: Concepts, Definitions, Formulas Refer to the listed pages to review the concepts, definitions, and formulas in this chapter that you need to understand. Section Pages 3.1 The definition of the derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 The derivative as a slope predictor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Differential notation for derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 Average and instantaneous rate of change of a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109–110, 114 Position function; velocity and acceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112–114 3.2 Operator notation for derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 The power rule: Dx x n = nx n−1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120, 126, 139 Linearity of differentiation: Dx (au + bv) = au + bv . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 The derivative of a polynomial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 The product rule: Dx (uv) = u v + uv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 The reciprocal rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 u v − uv u The quotient rule: Dx = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 v v2 dy dy du 3.3 The chain rule in differential notation: If y = u(x) then = · . . . . . . . . . . . . . 131 dx du d x The chain rule in functional notation: Dx f (g(x)) = f (g(x))g (x) . . . . . . . . . . . . . . . . . 132 du The generalized power rule: Dx u n = nu n−1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134, 140 dx 3.4 The definition of a vertical tangent line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 The continuity of differentiable functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 3.5 Maximum and minimum values of a function on a closed interval . . . . . . . . . . . . . . . . . . 147 The maximum-minimum value property for continuous functions . . . . . . . . . . . . . . . . . . 147 The necessary condition f (x) = 0 for a local extreme value . . . . . . . . . . . . . . . . . . . . . . . 148 Local and global (absolute) extreme values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148–149 The definition of a critical point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 The closed-interval maximum-minimum method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149–150 3.6 Steps in the solution of an applied maximum-minimum problem . . . . . . . . . . . . . . . . . . . 156 3.7 The sine-cosine derivatives: Dx sin x = cos x, Dx cos x = − sin x . . . . . . . . . . . . . . . . . . 170 The tangent-cotangent derivatives: Dx tan x = sec2 x, Dx cot x = − csc2 x . . . . . . . . . . 172 The secant-cosecant derivatives: Dx sec x = sec x tan x, Dx csc x = − csc x cot x . . . 172 Chain rule forms of the trigonometric differentiation formulas . . . . . . . . . . . . . . . . . . . . . 173 3.8 The general exponential function a x and the laws of exponents . . . . . . . . . . . . . . . . . . . . . 180–181 The number e ≈ 2.71828. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .184 The natural exponential function e x ; its derivative De x = e x . . . . . . . . . . . . . . . . . . . . . . . 184 The chain-rule exponential derivative: Dx eu = eu Dx u . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 The general logarithm function loga x and laws of logarithms . . . . . . . . . . . . . . . . . . . . . . 185, 187 Pairs of inverse functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 Differentiation of inverse functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 1 The natural logarithm function ln x; its derivative Dx ln x = . . . . . . . . . . . . . . . . . . . . . 188–189 x 1 du The chain-rule logarithmic derivative: Dx ln |u| = . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 u dx The process of logarithmic differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 3.9 Implicitly defined functions and implicit differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 194–195 Related-rates problems and derivatives of related functions . . . . . . . . . . . . . . . . . . . . . . . . 197 3.10 Iteration and the Babylonian square root method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205–206 Convergence of approximations to a solution of the equation f (x) = 0 . . . . . . . . . . . . . 207 f (xn ) . . . . . . . . . . . . . . . . . . . 207 The iterative formula of Newton’s method: xn+1 = xn − f (xn ) 218 Chapter 3 Miscellaneous Problems 219 CHAPTER 3: REVIEW (Continued) Objectives: Methods and Techniques Work the listed problems in each section to practice the methods and techniques in this chapter that you need to master. Section Problems 3.1 Using a differentiation rule to differentiate quadratic functions . . . . . . . . . . . . . . . . . . . . . 5, 9 Applying the definition of the derivative to find f (x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13, 17, 19 Finding when the velocity of a moving particle is zero . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25, 29, 39 Matching the graphs of a function and its derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31, 33 Calculating the rate of growth of a population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41, 53 Calculating rates of change in geometric situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45, 47, 49 3.2 Applying general differentiation rules to find derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 9, 11, 15, 19, 21, 27, 35 Finding tangent lines to graphs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43, 45, 49 Calculating rates of change in geometric situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51, 53 3.3 Using the chain rule to differentiate functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 9, 13, 15, 23, 25, 29, 33, 35 Calculating rates of change in geometric situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 51, 53, 57, 59 3.4 Using rules to differentiate algebraic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 9, 13, 17, 21, 23, 29, 35, 41 Finding tangent lines to graphs of algebraic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47, 49, 53 Matching the graphs of a function and its derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57, 59 3.5 Finding the maximum and minimum values of a function defined on a . . . . . . . . . . . . . . 5, 7, 11, 15, 19, 25, 33, closed interval 35, 37 Matching the graphs of a function and its derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 51 3.6 Solving applied maximum-minimum problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 7, 11, 17, 21, 23, 27, 31, 33, 45 3.7 Calculating derivatives of trigonometric functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5, 7, 9, 13, 15, 21, 27, 35, 39, 45, 47, 51, 53 Finding tangent lines to trigonometric graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61, 65 Solving trigonometric rate of change problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75, 77 Solving trigonometric maximum-minimum problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79, 81, 83 3.8 Calculating derivatives of exponential and logarithmic functions . . . . . . . . . . . . . . . . . . . 5, 7, 11, 15, 19, 23, 27, 29, 31, 33 Applying laws of logarithms before differentiating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39, 41 Finding a derivative by logarithmic differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49, 51 Finding tangent lines to exponential graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59, 61 3.9 Finding derivatives and tangent lines by implicit differentiation . . . . . . . . . . . . . . . . . . . . 3, 7, 15, 21, 25 Solving applied related-rates problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37, 39, 41, 43, 45, 47, 51, 53, 55, 61 3.10 Applying Newton’s method to find a solution of an equation . . . . . . . . . . . . . . . . . . . . . . . 3, 5, 9, 15, 17, 27, 33 MISCELLANEOUS PROBLEMS 1 11. y = 3 (x − x)3 Find dy/d x in Problems 1 through 35. 1. y = x 2 + 3. y = √ 3 x2 1 x+√ 3 x 2. y 2 = x 2 4. y = (x 2 + 4x)5/2 x4 + x2 +x +1 5. y = (x − 1)7 (3x + 2)9 6. y = 1 4 7. y = 3x − 2 2x 8. y = x sin 10x 9. x y = 9 x2 10 10. y = 1 5x 6 13. y = 1 1 + u2 where u = 14. x 3 = sin2 y √ 16. y = 3x 5 − 4x 2 17. y = u+1 , u−1 where u = 18. y = sin(2 cos 3x) √ 20. y = 1 + sin x 12. y = 1 1 + x2 15. y = √ √ √ 3 2x + 1 5 3x − 2 √ x+ 7/3 √ 3 2x x +1 19. x 2 y 2 = x + y √ 21. y = x + 2x + 3x 219 220 CHAPTER 3 The Derivative x + sin x x 2 + cos x 24. x 3 + y 3 = x y 22. y = 25. y = (1 + 2u) , 3 23. √ 3 x+ √ 3 y=4 Find the limits in Problems 64 through 69. x − tan x sin x sin 2x 66. lim x→0 sin 5x 1 68. lim x 2 sin 2 x→0 x 64. lim where 26. y = cos2 (sin2 x) x→0 1 u= (1 + x)3 27. y = sin2 x 1 + cos x √ 4 √ 3 28. y = 1 + x 1 − 2 3 x cos 2x 29. y = √ sin 3x 30. x 3 − x 2 y + x y 2 − y 3 = 4 31. y = e x cos x 33. y = [1 + (2 + 3e x )−3/2 ]2/3 32. y = e−2x sin 3x √ x −x 5 34. y = (e + e ) 35. y = cos3 3 1 + ln x Find the derivatives of the functions defined in Problems 36 through 45. 36. f (x) = cos(1 − e−x ) 38. f (x) = ln(x + e−x ) 37. f (x) = sin2 (e−x ) 39. f (x) = e x cos 2x 40. f (x) = e−2x sin 3x 41. g(t) = ln tet 42. g(t) = 3(et − ln t)5 2 + 3x 44. f (x) = e4x 43. g(t) = sin(et ) cos(e−t ) 1 + et 45. g(t) = 1 − et 2 47. sin(e x y ) = x 49. x = ye y 51. x ln y = x + y (x + 1)(x + 2) 1/3 (x 2 + 1)(x 2 + 2) √ √ √ 55. y = x + 1 3 x + 2 4 x + 3 x 57. y = (ln x)ln x , 56. y = x (e ) 80. 54. y = 81. x >1 In Problems 58 through 61, write an equation of the line tangent to the given curve at the indicated point. 58. 60. 61. 62. π (36x 2 − x 3 ) (in.3 ). 3 If the water flows out a hole at the bottom of the bowl at the rate of 36π in.3 /s, how fast is x decreasing when x = 6 in.? 63. Falling sand forms a conical sandpile. Its height h always remains twice its radius r while both are increasing. If sand is falling onto the pile at the rate of 25π ft3 /min, how fast is r increasing when r = 5 ft? 220 82. x +1 y= ; (0, −1) 59. x = sin 2y; (1, π/4) x −1 x 2 − 3x y + 2y 2 = 0; (2, 1) y 3 = x 2 + x; (0, 0) If a hemispherical bowl with radius 1 ft is filled with water to a depth of x in., then the volume of water in the bowl is V = 67. lim x 2 csc 2x cot 2x x→0 69. lim x→0+ h(x) = x + 79. In Problems 52 through 57, find dy/d x by logarithmic differentiation. √ (3 − x 2 )1/2 52. y = (x 2 − 4) 2x + 1 53. y = 4 (x + 1)1/4 x→0 √ x sin 1 x In Problems 70 through 75, identify two functions f and g such that h(x) = f (g(x)). Then apply the chain rule to find h (x). √ 1 71. h(x) = √ 70. h(x) = 3 x + x 4 x 2 + 25 x 72. h(x) = 73. h(x) = 3 (x − 1)5 2 x +1 (x + 1)10 74. h(x) = 75. h(x) = cos(x 2 + 1) (x − 1)10 76. The period T of oscillation (in seconds) of a √ simple pendulum of length L (in feet) is given by T = 2π L/32. What is the rate of change of T with respect to L when L = 4 ft? 77. What is the rate of change of the volume V = 43 πr 3 of a sphere with respect to its surface area A = 4πr 2 ? 78. What is an equation for the straight line through (1, 0) that is tangent to the graph of In Problems 46 through 51, find dy/d x by implicit differentiation. 46. xe y = y 48. e x + e y = e x y 50. e x−y = x y 65. lim x cot 3x 83. 84. 1 x at a point in the first quadrant? A rocket is launched vertically upward from a point 3 mi west of an observer on the ground. What is the speed of the rocket when the angle of elevation (from the horizontal) of the observer’s line of sight to the rocket is 60◦ and is increasing at 6◦ per second? An oil field containing 20 wells has been producing 4000 barrels of oil daily. For each new well drilled, the daily production of each well decreases by 5 barrels. How many new wells should be drilled to maximize the total daily production of the oil field? A triangle is inscribed in a circle of radius R. One side of the triangle coincides with a diameter of the circle. In terms of R, what is the maximum possible area of such a triangle? Five rectangular pieces of sheet metal measure 210 cm by 336 cm each. Equal squares are to be cut from all their corners, and the resulting five cross-shaped pieces of metal are to be folded and welded to form five boxes without tops. The 20 little squares that remain are to be assembled in groups of four into five larger squares, and these five larger squares are to be assembled into a cubical box with no top. What is the maximum possible total volume of the six boxes that are constructed in this way? A mass of clay of volume V is formed into two spheres. For what distribution of clay is the total surface area of the two spheres a maximum? A minimum? A right triangle has legs of lengths 3 m and 4 m. What is the maximum possible area of a rectangle inscribed in the triangle in the “obvious” way—with one corner at the triangle’s right angle, two adjacent sides of the rectangle lying on the triangle’s legs, and the opposite corner on the hypotenuse? Chapter 3 Miscellaneous Problems 85. What is the maximum possible volume of a right circular cone inscribed in a sphere of radius R? 86. A farmer has 400 ft of fencing with which to build a rectangular corral. He will use some or even all of an existing straight wall 100 ft long as part of the perimeter of the corral. What is the maximum area that can be enclosed? 87. In one simple model of the spread of a contagious disease among members of a population of M people, the incidence of the disease, measured as the number of new cases per day, is given in terms of the number x of individuals already infected by 221 93. Use the result of Problem 92 to show that the minimum distance from the point (x0 , y0 ) to a point of the straight line Ax + By + C = 0 is |Ax0 + By0 + C| . √ A2 + B 2 94. A race track is to be built in the shape of two parallel and equal straightaways connected by semicircles on each end (Fig. 3.MP.2). The length of the track, one lap, is to be exactly 4 km. What should its design be to maximize the rectangular area within it? R(x) = kx(M − x) = k M x − kx 2 , where k is a positive constant. How many individuals in the population are infected when the incidence R is the greatest? 88. Three sides of a trapezoid have length L, a constant. What should be the length of the fourth side if the trapezoid is to have maximal area? 89. A box with no top must have a base twice as long as it is wide, and the total surface area of the box is to be 54 ft2 . What is the maximum possible volume of such a box? 90. A small right circular cone is inscribed in a larger one (Fig. 3.MP.1). The larger cone has fixed radius R and fixed altitude H . What is the largest fraction of the volume of the larger cone that the smaller one can occupy? R H FIGURE 3.MP.1 A small cone inscribed in a larger one (Problem 90). 91. Two vertices of a trapezoid are at (−2, 0) and (2, 0), and the other two lie on the semicircle x 2 + y 2 = 4, y 0. What is the maximum possible area of the trapezoid? [Note: The area of a trapezoid with bases b1 and b2 and height h is A = h(b1 + b2 )/2.] 92. Suppose that f is a differentiable function defined on the whole real number line R and that the graph of f contains a point Q(x, y) closest to the point P(x0 , y0 ) not on the graph. Show that x − x0 f (x) = − y − y0 at Q. Conclude that the segment PQ is perpendicular to the line tangent to the curve at Q. [Suggestion: Minimize the square of the distance PQ.] FIGURE 3.MP.2 Design the race track to maximize the rectangular area (Problem 94). 95. Two towns are located near the straight shore of a lake. Their nearest distances to points on the shore are 1 mi and 2 mi, respectively, and these points on the shore are 6 mi apart. Where should a fishing pier be located to minimize the total amount of paving necessary to build a straight road from each town to the pier? 96. A hiker finds herself in a forest 2 km from a long straight road. She wants to walk to her cabin, which is 10 km away in the forest and also 2 km from the road (Fig. 3.MP.3). She can walk at a rate of 8 km/h along the road but only 3 km/h through the forest. So she decides to walk first to the road, then along the road, and finally through the forest to the cabin. What angle θ (shown in the figure) would minimize the total time required for the hiker to reach her cabin? How much time is saved in comparison with the straight route through the forest? 10 Hiker 2 θ Cabin Forest θ Road FIGURE 3.MP.3 The hiker’s quickest path to the cabin (Problem 96). 97. When an arrow is shot from the origin with initial velocity v and initial angle of inclination α (from the horizontal x-axis, which represents the ground), then its trajectory is the curve y = mx − 16 (1 + m 2 )x 2 , v2 where m = tan α. (a) Find the maximum height reached by the arrow in terms of m and v. (b) For what value of m (and hence, for what α) does the arrow travel the greatest horizontal distance? 221 222 CHAPTER 3 The Derivative 98. A projectile is fired with initial velocity v and angle of elevation θ from the base of a plane inclined at 45◦ from the horizontal (Fig. 3.MP.4). The range of the projectile, as measured up this slope, is given by √ v2 2 R= (cos θ sin θ − cos2 θ). 16 What value of θ maximizes R? θ 115. Criticize the following “proof” that 3 = 2. Begin by writing x3 = x · x2 = x2 + x2 + · · · + x2 (x summands). Differentiate to obtain 3x 2 = 2x + 2x + · · · + 2x (still x summands). Thus 3x 2 = 2x 2 , and “therefore” 3 = 2. 45° R Level Ground If we substitute z = x + h into the definition of the derivative, the result is f (z) − f (x) . f (x) = lim z→x z−x Use this formula in Problems 116 and 117, together with the formula a 3 − b3 = (a − b)(a 2 + ab + b2 ) for factoring the difference of two cubes. FIGURE 3.MP.4 A projectile fired uphill (Problem 98). In Problems 99 through 110, use Newton’s method to find the solution of the given equation f (x) = 0 in the indicated interval [a, b] accurate to four decimal places. 99. x 2 −7 = 0; [2, 3] 100. x − 3 = 0; [1, 2] 3 101. x 5 − 75 = 0; (to find the positive square root of 7) (to find the fifth root of 75) [5, 6] (to approximate 103/4 ) 103. x 3 − 3x − 1 = 0; [−1, 0] 104. x 3 − 4x − 1 = 0; [−1, 0] 105. e −x − sin x = 0; [0, 2] 107. x + cos x = 0; [−2, 0] 108. x 2 + sin x = 0; [−1.0, −0.5] [Suggestion: Factor the numerator as a difference of cubes and the denominator as a difference of squares.] 117. Prove that 109. 4x − sin x + 4 = 0; [−2, −1] 110. 5x − cos x + 5 = 0; [−1, 0] 112. The equation x 2 + 1 = 0 has no real solutions. Try finding a solution by using Newton’s method and report what happens. Use the initial estimate x0 = 2. 113. At the beginning of Section 3.10 we mentioned the fifthdegree equation x 5 − 3x 3 + x 2 − 23x + 19 = 0; its graph appears in Fig. 3.10.1. The graph makes it clear that this equation has exactly three real solutions. Find all of them, to four-place accuracy, using Newton’s method. 114. The equation tan x = Dx x 2/3 = lim z→x 2 z 2/3 − x 2/3 = x −1/3 . z−x 3 [Suggestion: Factor the numerator as a difference of squares and the denominator as a difference of cubes.] 119. Air is being pumped into a spherical balloon at the constant rate of 10 in.3 /s. At what rate is the surface area of the balloon increasing when its radius is 5 in.? 111. Find the depth to which a wooden ball with radius 2 ft sinks in water if its density is one-third that of water. A useful formula appears in Problem 37 of Section 3.10. 1 x has a sequence α1 , α2 , α3 , . . . of positive roots, with αn slightly larger than (n − 1)π. Use Newton’s method to approximate α1 and α2 to three-place accuracy. 222 z→x 3 z 3/2 − x 3/2 = x 1/2 . z−x 2 118. A rectangular block with square base is being squeezed in such a way that its height y is decreasing at the rate of 2 cm/min while its volume remains constant. At what rate is the edge x of its base increasing when x = 30 cm and y = 20 cm? [0, 2] 106. cos x − ln x = 0; Dx x 3/2 = lim (to find the cube root of 3) [2, 3] 102. x 4/3 − 10 = 0; 116. Show that 120. A ladder 10 ft long is leaning against a wall. If the bottom of the ladder slides away from the wall at the constant rate of 1 mi/h, how fast (in miles per hour) is the top of the ladder moving when it is 0.01 ft above the ground? 121. A water tank in the shape of an inverted cone, axis vertical and vertex downward, has a top radius of 5 ft and height 10 ft. Water is flowing out of the tank through a hole at the vertex at the rate of 50 ft3 /min. What is the time rate of change of the water depth at the instant when the water is 6 ft deep? 122. Plane A is flying west toward an airport at an altitude of 2 mi. Plane B is flying south toward the same airport at an altitude of 3 mi. When both planes are 2 mi (ground distance) from the airport, the speed of plane A is 500 mi/h and the distance between the two planes is decreasing at 600 mi/h. What is the speed of plane B then? 123. A water tank is shaped in such a way that the volume of water in the tank is V = 2y 3/2 in.3 when its depth is y inches. If water flows out through a hole at the bottom of the tank √ at the rate of 3 y in.3 /min, at what rate does the water level in the tank fall? What is a practical application for such a water tank? Chapter 3 Miscellaneous Problems 124. Water is being poured into the conical tank of Problem 121 at the rate of 50 ft3 /min and is draining through the hole at √ the bottom at the rate of 10 y ft3 /min, where y is the depth of water in the tank. (a) At what rate is the water level rising when the water is 5 ft deep? (b) Suppose that the tank is initially empty, water is poured in at 25 ft3 /min, and water √ continues to drain at 10 y ft3 /min. What is the maximum depth attained by the water? 223 125. Let L be a straight line passing through the fixed point P(x0 , y0 ) and tangent to the parabola y = x 2 at the point Q(a, a 2 ). (a) Show that a 2 − 2ax0 + y0 = 0. (b) Apply the quadratic formula to show that if y0 < (x0 )2 (that is, if P lies below the parabola), then there are two possible values for a and thus two lines through P that are tangent to the parabola. (c) Similarly, show that if y0 > (x0 )2 (P lies above the parabola), then no line through P can be tangent to the parabola. PHOTO CREDITS p. 105 Ward’s Natural Science Establishment/Science Source; (bottom right) C. H. Edwards p. 168 Richard Megna/Fundamental Photographs 223 This page intentionally left blank 4 Additional Applications of the Derivative G ottfried Wilhelm Leibniz entered the University of Leipzig when he was 15, studied philosophy and law, graduated at 17, and received his doctorate in philosophy at 21. Upon completion of his academic work, Leibniz entered the political and governmental service of the Elector of Mainz (Germany). His seG. W. Leibniz (1646–1716) rious study of mathematics did not begin until 1672 (when he was 26) when he was sent to Paris on a diplomatic mission. During the next four years there he conceived the principal features of calculus. For this work he is remembered (with Newton) as a codiscoverer of the subject. Newton’s discoveries had come slightly earlier (in the late 1660s), but Leibniz’s were the first to be published, beginning in 1684. Despite an unfortunate priority dispute between supporters of Newton and supporters of Leibniz that raged for more than a century, it is clear now that the discoveries were made independently. Throughout his life, Leibniz sought a universal language incorporating notation and terminology that would provide all educated people with the powers of clear and correct reasoning in all subjects. But only in mathematics did he largely accomplish this goal. His differential notation for calculus is arguably the best example of a system of notation chosen so as to mirror perfectly the basic operations and processes of the subject. Indeed, it can be said that Leibniz’s notation for calculus brings within the range of ordinary students problems that once required the ingenuity of an Archimedes or a Newton. For this reason, Leibniz’s approach to calculus was dominant during the eighteenth century, even though Newton’s somewhat different approach may have been closer to our modern understanding of the subject. The origin of differential notation was an infinitesimal right triangle with legs d x and dy and with hypotenuse a tiny segment of the curve y = f (x). Leibniz later described the moment he first visualized this “characteristic triangle” as a burst of light that was the inception of his calculus. Indeed, he sometimes referred to his calculus as “my method of the Characteristic Triangle.” y = f (x) dy dx Leibniz’s characteristic triangle The following excerpt shows the opening paragraphs of Leibniz’s first published article (in the 1684 Acta Eruditorum) in which the differential notation initially appeared. In the fifth line of the second paragraph, the product rule for differentiation is expressed as d(xv) = x dv + v d x. From Chapter 4 of Calculus, Early Transcendentals, Seventh Edition. C. Henry Edwards, David E. Penney. Copyright © 2008 by Pearson Education, Inc. All rights reserved. 225 226 CHAPTER 4 Additional Applications of the Derivative 4.1 INTRODUCTION We learned in Chapter 3 how to differentiate a wide variety of algebraic and trigonometric functions. We saw that derivatives have such diverse applications as maximumminimum problems, related-rates problems, and the solution of equations by Newton’s method. The further applications of differentiation that we discuss in this chapter all depend ultimately upon a single fundamental question. Suppose that y = f (x) is a differentiable function defined on the closed interval [a, b ] of length x = b − a. Then the increment y in the value of f (x) as x changes from x = a to x = b = a + x is y = f (b) − f (a). (1) The question is this: How is the increment y related to the derivative—the rate of change—of the function f at the points of the interval [a, b ]? An approximate answer is given in Section 4.2. If the function continued throughout the interval with the same rate of change f (a) that it had at x = a, then the change in its value would be f (a)(b − a) = f (a) x. This observation motivates the tentative approximation y ≈ f (a) x. (2) A precise answer to the preceding question is provided by the mean value theorem of Section 4.3. This theorem implies that the exact increment is given by y = f (c) x (3) for some number c in (a, b). The mean value theorem is the central theoretical result of differential calculus, and is also the key to many of the more advanced applications of derivatives. 4.2 INCREMENTS, DIFFERENTIALS, AND LINEAR APPROXIMATION Sometimes we need a quick and simple estimate of the change in f (x) that results from a given change in x. We write y for f (x) and suppose first that the change in the independent variable is the increment x, so that x changes from its original value to the new value x + x. The change in the value of y is the increment y, computed by subtracting the old value of y from its new value: y y = f(x) f (x + Δ x) Δy y = f (x + x) − f (x). (1) f (x) Δx x x + Δx FIGURE 4.2.1 The increments x and y. x The increments x and y are represented geometrically in Fig. 4.2.1. Now we compare the actual increment y with the change that would occur in the value of y if it continued to change at the fixed rate f (x) while the value of the independent variable changes from x to x + x. This hypothetical change in y is the differential dy = f (x) x. (2) As Fig. 4.2.2 shows, dy is the change in height of a point that moves along the tangent line at the point (x, f (x)) rather than along the curve y = f (x). Think of x as fixed. Then Eq. (2) shows that the differential dy is a linear function of the increment x. For this reason, dy is called the linear approximation to the increment y. We can approximate f (x + x) by substituting dy for y: f (x + x) = y + y ≈ y + dy. Because y = f (x) and dy = f (x)x, this gives the linear approximation formula f (x + x) ≈ f (x) + f (x) x. 226 (3) Increments, Differentials, and Linear Approximation SECTION 4.2 227 y y = f(x) Tangent line at (x, f (x)) y + Δy y + dy dy Δy y Δx x x + Δx x FIGURE 4.2.2 The estimate dy of the actual increment y. The point is that this approximation is a “good” one, at least when x is relatively small. If we combine Eqs. (1), (2), and (3), we see that y ≈ f (x) x = dy. (4) Thus the differential dy = f (x) x is a good approximation to the increment y = f (x + x) − f (x). If we replace x with a in Eq. (3), we get the approximation y y = f (x) f (a + x) ≈ f (a) + f (a) x. (5) If we now write x = x − a, so that x = a + x, the result is (a, f (a)) y = f(a) + f'(a)(x − a) f (x) ≈ f (a) + f (a) · (x − a). (6) Because the right-hand side x a L(x) = f (a) + f (a) · (x − a) FIGURE 4.2.3 The graph of the linear approximation L(x) = f (a) + f (a) · (x − a) is the line tangent to y = f (x) at the point (a, f (a)). 2 1.8 1.6 1.4 1.2 (0, 1) y 1 0.8 0.6 0.4 y = 1 + x 0.2 0 −0.5 0 y=1+ 1 2 dy Δy in Eq. (6) is a linear function of x, we call it the linear approximation L(x) to the function f (x) near the point x = a. As illustrated in Fig. 4.2.3, the graph y = L(x) is the straight line tangent to the graph y = f (x) at the point (a, f (a)). EXAMPLE 1 Find the linear approximation to the function f (x) = point a = 0. x √ 1 + x near the Solution Note that f (0) = 1 and that f (x) = Δx (7) 1 1 (1 + x)−1/2 = √ , 2 2 1+x so f (0) = 12 . Hence Eq. (6) with a = 0 yields x 0.5 1 FIGURE√4.2.4 The function f (x) = 1 + x and its linear approximation L(x) = 1 + 12 x near a = 0. f (x) ≈ f (0) + f (0) · (x − 0) = 1 + 12 x = L(x). Thus the desired linear approximation is √ 1 + x ≈ 1 + 12 x. (8) Figure 4.2.4√illustrates the close approximation near x = 0 of the nonlinear ◗ function f (x) = 1 + x by its linear approximation L(x) = 1 + 12 x. 227 228 CHAPTER 4 Additional Applications of the Derivative IMPORTANT It is evident in Fig. 4.2.4 that the value of the linear √ approximation L(x) = 1 + 12 x is closer to the actual value of the function f (x) = 1 + x when x is closer to a = 0. For instance, the approximate values √ 1.1 ≈ 1 + 12 (0.1) = 1.05 (using x = 0.1 in (8)) and √ 1.03 ≈ 1 + 12 (0.03) = 1.015 (using x = 0.03 in (8)) are accurate to two and three decimal places (rounded), respectively. But √ 3 ≈ 1 + 12 · 2 = 2, √ using x = 2, is a very poor approximation to 3 ≈ 1.732. √ The approximation 1 + x ≈ 1 + 12 x is a special case of the approximation (1 + x)k ≈ 1 + kx (9) (k is a constant, x is near zero), an approximation with numerous applications. The derivation of (9) is similar to that in Example 1. (See Problem 39.) EXAMPLE 2 Use the linear approximation formula to approximate (122)2/3 . Note that 2 (125)2/3 = (125)1/3 = 52 = 25. Solution We need to approximate a particular value of x 2/3 , so our strategy is to apply Eq. (6) with f (x) = x 2/3 . We first note that f (x) = 23 x −1/3 . We choose a = 125, because we know the exact values f (125) = (125)2/3 = 25 and f (125) = 23 (125)−1/3 = 2 15 and because 125 is relatively close to 122. Then the linear approximation in (6) to f (x) = x 2/3 near a = 125 takes the form f (x) ≈ f (125) + f (125) · (x − 125); that is, x 2/3 ≈ 25 + 2 (x 15 − 125). With x = 122 we get (122)2/3 ≈ 25 + 2 (−3) 15 = 24.6. Thus (122)2/3 is approximately 24.6. The actual value of (122)2/3 is about 24.5984, so ◗ the formula in (6) gives a relatively good approximation in this case. Δx 10 x EXAMPLE 3 A hemispherical bowl of radius 10 in. is filled with water to a depth of x inches. The volume V of water in the bowl (in cubic inches) is given by the formula π V = (30x 2 − x 3 ) (10) 3 (Fig. 4.2.5). (You will be able to derive this formula after you study Chapter 6.) Suppose that you measure the depth of water in the bowl to be 5 in. with a maximum 1 in. Estimate the maximum error in the calculated volpossible measured error of 16 ume of water in the bowl. Solution The error in the calculated volume V (5) is the difference FIGURE 4.2.5 The bowl of Example 3. V = V (x) − V (5) between the actual volume V (x) and the calculated volume. We do not know the depth x of water in the bowl. We are given only that the difference x = x − 5 228 Increments, Differentials, and Linear Approximation SECTION 4.2 229 between the actual and the measured depths is numerically at most Because Eq. (10) yields V (x) = 1 16 in.: |x| 1 . 16 π (60x − 3x 2 ) = π(20x − x 2 ), 3 the linear approximation V ≈ d V = V (5) x at x = 5 gives V ≈ π(20 · 5 − 52 ) x = 75π x. 1 1 With the common practice in science of writing x = ± 16 to signify that − 16 1 x 16 , this gives V ≈ (75π ) ± 1 16 ≈ ±14.73 (in.3 ). The formula in Eq. (10) gives the calculated volume V (5) ≈ 654.50 in.3 , but we now ◗ see that this may be in error by almost 15 in.3 in either direction. Absolute and Relative Errors The (absolute) error in a measured or approximated value is defined to be the remainder when the approximate value is subtracted from the true value. Hence “actual value = approximate value + error.” The relative error is the ratio of the (absolute) error to the true value, “relative error = error ,” value and may be given as either a numerical fraction or as a percentage of the value. EXAMPLE 4 In Example 3, a relative error in the measured depth x of 1 x = 16 = 0.0125 = 1.25% x 5 leads to a relative error in the estimated volume of dV 14.73 ≈ ≈ 0.0225 = 2.25%. V 654.50 The relationship between these two relative errors is of some interest. The formulas for d V and V in Example 3 give 3(20 − x) x π(20x − x 2 ) x dV = = 1 · . 2 3 V 30 − x x π(30x − x ) 3 When x = 5, this gives dV x = (1.80) . V x Hence, to approximate the volume of water in the bowl with a relative error of at most 0.5%, for instance, we would need to measure the depth with a relative error of at most ◗ (0.5%)/1.8, thus with a relative error of less than 0.3%. 229 230 CHAPTER 4 Additional Applications of the Derivative The Error in Linear Approximation Now we consider briefly the question of the difference between the values of a function f (x) and its linear approximation L(x) near the point x = a. If we let x = x − a and write y = f (x), f (a + x) = f (a) + y, and L(x) = f (a) + f (a) · x = f (a) + dy, it then follows that the error in the linear approximation is given by f (x) − L(x) = y − dy, as illustrated in Fig. 4.2.6. It appears in the figure that, the smaller x is, the closer are the corresponding points on the curve y = f (x) and its tangent line y = L(x). Because Eq. (11) implies that the difference in the heights of two such points is equal to y − dy, the figure suggests that y − dy approaches zero as x → 0. But even more is true: The difference y error y − dy = f (a + x) − f (a) − f (a) x Δy dy = f '(a) Δx (x) = a + Δx (12) is a function of x that is small even in comparison with x. To see why, let’s write Δx a (11) x y − dy f (a + x) − f (a) = − f (a) x x and note that FIGURE 4.2.6 The error y − dy in the linear approximation dy ≈ f (a) x = dy. lim (x) = f (a) − f (a) = 0. x→0 Consequently, the error y − dy = (x) · x (13) in the linear approximation dy = f (a) x to the actual increment y is the product of two quantities, both of which approach zero as x → 0. If x is “very small”—so that (x) is also “very small”—then we might well describe their product in (13) as “very very small.” In this case we may finally rewrite Eq. (13) in the form f (a + x) − f (a) = f (a) x + (x) · x, (14) expressing the actual increment y = f (a + x) − f (a) as the sum of the very small differential dy = f (a) x and the very very small error (x)·x in this differential. EXAMPLE 5 If y = f (x) = x 3 , then simple computations (with x = x − a) give y = f (a + x) − f (a) = (a + x)3 − a 3 = 3a 2 x + 3a(x)2 + (x)3 and dy = f (a) x = 3a 2 x. Hence y − dy = 3a(x)2 + (x)3 . If a = 1 and x = 0.1, for instance, then these formulas yield y = 0.331, dy = 0.3, and y − dy = 0.031, thereby illustrating the smallness in the error y − dy in the linear approximation in comparison with the values of y and dy. ◗ 230 Increments, Differentials, and Linear Approximation SECTION 4.2 231 Example 6 indicates how we sometimes can use a graphing calculator or computer to specify how accurate a linear approximation is—in terms of its accuracy throughout an entire interval containing the point x = a. In concrete situations we often want to determine an interval throughout which the linear approximation provides a specified accuracy. 2.5 y= 2 1.5 y 1 + x + 0.1 (0, 1) y=1+ 0.5 y= − 0.5 0 1 x 2 0.5 1 1.5 x 1.6 y= 1 + x + 0.1 1.4 1.2 Solution Accuracy to within 0.1 means that the two functions in (15) differ by less than 0.1: 1 + x − 0.1 FIGURE√4.2.7 The function f (x) = 1 + x on the interval −1 < x < 1.5. y (15) of Example 1 is accurate to within 0.1. 1 0 −1 EXAMPLE 6 Find an interval on which the approximation √ 1 + x ≈ 1 + 12 x √ 1 + x − 1 + 1 x < 0.1, 2 which is equivalent to √ √ 1 + x − 0.1 < 1 + 12 x < 1 + x + 0.1. Thus we want the graph of the linear approximation y = 1 + 12 x to lie between the √ two curves obtained by shifting the graph y = 1 + x vertically up and down by the amount 0.1. Figure 4.2.7 shows the graphs of all these curves on the interval −1 < x < 1.5. The points at which the linear approximation y = 1 + 12 x emerges √ from the band of width 0.2 around the graph y = 1 + x are marked, and we see that a smaller interval around x = 0 is needed to confine the linear approximation within the desired range. Indeed, the zoom shown in Fig. 4.2.8 indicates that the approximation in (15) is accurate to within 0.1 for every x in the interval −0.6 < x < 0.9. ◗ (0, 1) y=1+ 1 1 x 2 0.8 y= 0.6 1 + x − 0.1 Differentials The linear approximation formula in (3) is often written with d x in place of x: f (x + d x) ≈ f (x) + f (x) d x. 0.4 − 0.4 − 0.2 0 0.2 0.4 0.6 0.8 x FIGURE√4.2.8 The function f (x) = 1 + x on the smaller interval −0.6 < x < 0.9. (16) In this case d x is an independent variable, called the differential of x, and x is fixed. Thus the differentials of x and y are defined to be d x = x and dy = f (x) x = f (x) d x. (17) From this definition it follows immediately that dy dx FIGURE 4.2.9 The slope of the tangent line as the ratio of the infinitesimals dy and d x. f (x) d x dy = = f (x), dx dx in perfect accord with the notation we have been using. Indeed, Leibniz originated differential notation by visualizing “infinitesimal” increments d x and dy (Fig. 4.2.9), with their ratio dy/d x being the slope of the tangent line. The key to Leibniz’s independent discovery of differential calculus in the 1670s was his insight that if d x and dy are sufficiently small, then the segment of the curve y = f (x) and the straight line segment joining (x, y) and (x + d x, y + dy) are virtually indistinguishable. This insight is illustrated by the successive magnifications in Figs. 4.2.10 through 4.2.12 of the curve y = x 2 near the point (1, 1). Differential notation provides us with a convenient way to write derivative formulas. Suppose that z = f (u), so that dz = f (u) du. For particular choices of the function f , we get the formulas d(u n ) = nu n−1 du, d(sin u) = (cos u) du, d(eu ) = eu du, 231 232 CHAPTER 4 Additional Applications of the Derivative 8 1.4 2.5 y= 6 y = x2 y = x2 x2 2 y 1.2 y y 1.5 4 1 dy 2 1 dx 0 0 1 dy dy 2 0.5 3 x FIGURE 4.2.10 d x = 1. dx dx 0.8 1 1.2 x 1.4 0.9 1.6 1 1.1 1.2 x FIGURE 4.2.11 d x = 13 . FIGURE 4.2.12 d x = 1 10 . and so on. Thus we can write differentiation rules in differential form without having to identify the independent variable. The sum, product, and quotient rules take the respective forms d(u + v) = du + dv, d(uv) = u dv + v du, and u v du − u dv . v v2 If z = f (u) and u = g(x), we may substitute du = g (x) d x into the formula dz = f (u) du. This gives d = dz = f (g(x)) · g (x) d x. This is the differential form of the chain rule Dx f (g(x)) = f (g(x)) · g (x). Thus the chain rule appears here as though it were the result of mechanical manipulations of the differential notation. This compatibility with the chain rule is one reason for the extraordinary usefulness of differential notation in calculus. EXAMPLE 7 √ (a) If y = 3x 2 − 2x 3/2 , then dy = 6x − 3 x d x. (b) If u = sin2 t − cos 2t, then du = (2 sin t cos t + 2 sin 2t) dt = 3 sin 2t dt (using the trigonometric identity sin 2t = 2 sin t cos t). (c) If w = ze z , then dw = (1 · e z + z · e z ) dz = (1 + z)e z dz. ◗ Proof of the Chain Rule We can now use our knowledge of the error in linear approximations to give a proof of the chain rule for the composition f ◦ g that does not require the assumption g (x) = 0 that we needed in Section 3.3. Here we suppose only the existence of the derivatives g (a) and f (b) (where b = g(a)) of the functions u = g(x) and y = f (g(x)) = f (u). If we write u = g(a + x) − g(a) and y = f (b + u) − f (b), then Eq. (14) in this section—with g in place of f —gives u = g (a) x + 1 · x = g (a) + 1 x 232 (18) Increments, Differentials, and Linear Approximation SECTION 4.2 233 where 1 → 0 as x → 0. A second application of Eq. (14)—this time with u in place of x—gives y = f (b) u + 2 · u = f (b) + 2 u = f (g(a)) + 2 · g (a) + 1 x (19) where 2 → 0 as u → 0, and hence as x → 0 (because Eq. (18) shows that u → 0 as x → 0). Finally, when we divide by x in Eq. (19) and then take the limit as x → 0, we get y dy = lim = lim f (g(a)) + 2 · g (a) + 1 = f (g(a)) · g (a). x→0 x x→0 dx Thus we have shown that the chain rule formula Dx [ f (g(x))] = f (g(x)) · g (x) holds at x = a. 4.2 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Suppose that y = f (x) and that x is an increment in x. Then, by definition, y = f (x + x) − f (x). 2. If y = f (x) and x is an increment in x, then, by definition, dy = f (x)x. 3. If y = f (x) and x is an increment in x, then the linear approximation formula states that f (x + x) ≈ f (x) + f (x)x. √ √ 4. The linear approximation to f (x) = 1 + x near the point a = 0 is 1 + x ≈ 1 + x. 5. In Example 2 we find that (122)2/3 = 24.6. 6. The error in the linear approximation L(x) = f (x) + f (a)x to the function f near the point x = a is f (x) − L(x) = y − dy. 7. d(u n ) = nu n−1 . 8. d(sin u) = (cos u) du. 9. d(uv) = u dv + v du. 10. If w = w(z) = ze z , then dw = (1 + z)e z dz. 4.2 CONCEPTS: QUESTIONS AND DISCUSSION 1. Use Eqs. (11)–(13) of this section to show that the linear function L(x) = f (a)+ f (a) · (x − a) of x satisfies the condition f (x) − L(x) = 0. (20) x→a x −a 2. Any linear function L(x) = mx + b satisfying Eq. (20) is called a linearization of the function f (x) at the point x = a. Can a function have two different linearizations at the same point? 3. Can a function have a linearization (as in Question 2) at a point where it is not differentiable? lim 4.2 PROBLEMS In Problems 1 through 16, write dy in terms of x and d x. 1. y = 3x 2 − 3. y = x − √ 4 x2 √ 3 2. y = 2 x − √ 3 x 4 − x3 4. y = 1 √ x− x 5. y = 3x 2 (x − 3)3/2 7. y = x(x 2 + 25)1/4 9. y = cos √ x 11. y = sin 2x cos 2x x x2 − 4 1 8. y = 2 (x − 1)4/3 6. y = 10. y = x 2 sin x 12. y = cos3 3x 233 234 CHAPTER 4 Additional Applications of the Derivative sin 2x 3x 1 15. y = 1 − x sin x 14. y = x 3 e−2x 13. y = h ln x 16. y = x r In Problems 17 through 24, find—as in Example 1—the linear approximation L(x) to the given function f (x) near the point a = 0. 1 1 18. f (x) = √ 17. f (x) = 1−x 1+x 19. f (x) = (1 + x)2 20. f (x) = (1 − x)3 21. f (x) = (1 − 2x)3/2 22. f (x) = e−x 23. f (x) = sin x 24. f (x) = ln(1 + x) In Problems 25 through 34, use—as in Example 2—a linear approximation L(x) to an appropriate function f (x), with an appropriate value of a, to estimate the given number. √ √ 25. 3 25 26. 102 √ √ 28. 80 27. 4 15 29. 65−2/3 30. 803/4 31. cos 43◦ 32. sin 32◦ 34. ln 11 10 33. e1/10 In Problems 35 through 38, compute the differential of each side of the given equation, regarding x and y as dependent variables (as if both were functions of some third, unspecified, variable). Then solve for dy/d x. 35. x 2 + y 2 = 1 36. xe y = 1 37. x 3 + y 3 = 3x y 38. x ln y = 1 for any real constant k (which 39. Assuming that Dx x = kx we shall establish in Chapter 6), derive the linear approximation formula (1 + x)k ≈ 1 + kx for x near zero. k k−1 In Problems 40 through 47, use linear approximations to estimate the change in the given quantity. FIGURE 4.2.15 The conical sandpile of Problem 44— volume V = 13 πr 2 h. 1 2 45. The range R = 32 v sin 2θ of a shell fired at inclination ◦ angle θ = 45 , if its initial velocity v is increased from 80 ft/s to 81 ft/s. 1 2 v sin 2θ of a projectile fired with initial 46. The range R = 32 velocity v = 80 ft/s, if its initial inclination angle θ is increased from 60◦ to 61◦ . 47. The wattage W = R I 2 of a floodlight with resistance R = 10 ohms, if the current I is increased from 3 amperes to 3.1 amperes. 48. The equatorial radius of the earth is approximately 3960 mi. Suppose that a wire is wrapped tightly around the earth at the equator. Approximately how much must this wire be lengthened if it is to be strung all the way around the earth on poles 10 ft above the ground? Use the linear approximation formula! 49. The radius of a spherical ball is measured as 10 in., with a maximum error of 161 in. What is the maximum resulting error in its calculated volume? 50. With what accuracy must the radius of the ball of Problem 49 be measured to ensure an error of at most 1 in.3 in its calculated volume? 51. The radius of a hemispherical dome is measured as 100 m with a maximum error of 1 cm (Fig. 4.2.16). What is the maximum resulting error in its calculated surface area? r 40. The circumference of a circle, if its radius is increased from 10 in. to 10.5 in. 41. The area of a square, if its edge length is decreased from 10 in. to 9.8 in. 42. The surface area of a sphere, if its radius is increased from 5 in. to 5.2 in. (Fig. 4.2.13). r r FIGURE 4.2.13 The sphere of Problem 42—area A = 4πr 2 , volume V = 43 πr 3 . h FIGURE 4.2.14 The cylinder of Problem 43—volume V = πr 2 h. 43. The volume of a cylinder, if both its height and its radius are decreased from 15 cm to 14.7 cm (Fig. 4.2.14). 44. The volume of the conical sandpile of Fig. 4.2.15, if its radius is 14 in. and its height is increased from 7 in. to 7.1 in. 234 FIGURE 4.2.16 The hemisphere of Problem 51—curved surface area A = 2πr 2 . 52. With what accuracy must the radius of a hemispherical dome be measured to ensure an error of at most 0.01% in its calculated surface area? In Problems 53 through 60, a function f (x) and a point x = a are given. Determine graphically an open interval I centered at a so that the function f (x) and its linear approximation L(x) differ by less than the given value at each point of I . 53. f (x) = x 2 , a = 1, = 0.2 √ 54. f (x) = x, a = 1, = 0.1 1 55. f (x) = , a = 2, = 0.01 x √ 56. f (x) = 3 x, a = 8, = 0.01 57. f (x) = sin x, a = 0, = 0.05 58. f (x) = e x , a = 0, = 0.05 59. f (x) = sin x, a = π/4, = 0.02 60. f (x) = tan x, a = π/4, = 0.02 Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 235 4.3 INCREASING AND DECREASING FUNCTIONS AND THE MEAN VALUE THEOREM The significance of the sign of the first derivative of a function is simple but crucial: f (x) is increasing on an interval where f (x) > 0; f (x) is decreasing on an interval where f (x) < 0. Geometrically, this means that where f (x) > 0, the graph of y = f (x) is rising as you scan it from left to right. Where f (x) < 0, the graph is falling. We can clarify the terms increasing and decreasing as follows. DEFINITION Increasing and Decreasing Functions The function f is increasing on the interval I = (a, b) provided that f (x1 ) < f (x2 ) for all pairs of numbers x1 and x2 in I for which x1 < x2 . The function f is decreasing on I provided that f (x1 ) > f (x2 ) for all pairs of numbers x1 and x2 for which x1 < x2 . Figure 4.3.1 illustrates this definition. In short, the function f is increasing on I = (a, b) if the values of f (x) increase as x increases [Fig. 4.3.1(a)]; f is decreasing on I if the values of f (x) decrease as x increases [Fig. 4.3.1(b)]. f increasing on (a, b) f(x2 ) f(x1) a x1 y x2 b 4 f decreasing on (a, b) 3 (a) y = x2 2 f(x1) a x1 f(x2 ) x2 1 b (b) FIGURE 4.3.1 (a) An increasing function and (b) a decreasing function. −2 −1 1 2 x FIGURE 4.3.2 f (x) = x 2 is decreasing for x < 0, increasing for x > 0. EXAMPLE 1 As illustrated in Fig. 4.3.2, the simple function f (x) = x 2 is decreasing on the interval (−∞, 0) and increasing on the interval (0, +∞). This follows immediately from the elementary fact that u 2 < v 2 if 0 < u < v. Because f (x) = 2x, we also see immediately that f (x) < 0 on the interval (−∞, 0) and that f (x) > 0 on the interval (0, +∞). But for more general functions, the mean value theorem of this section is needed to establish the precise relationship between the sign of the derivative of a function and its increasing-decreasing behavior. ◗ REMARK We speak of a function as increasing or decreasing on an interval, not at a single point. Nevertheless, if we consider the sign of f , the derivative of f , at a single point, we get a useful intuitive picture of the significance of the sign of the 235 236 CHAPTER 4 Additional Applications of the Derivative derivative. This is because the derivative f (x) is the slope of the tangent line at the point (x, f (x)) on the graph of f . If f (x) > 0, then the tangent line has positive slope. Therefore, it rises as you scan from left to right. Intuitively, a rising tangent would seem to correspond to a rising graph and thus to an increasing function. Similarly, we expect to see a falling graph where f (x) is negative (Fig. 4.3.3). One caution: In order to determine whether a function f is increasing or decreasing, we must examine the sign of f on a whole interval, not merely at a single point. (See Problem 59.) The Mean Value Theorem x f '(x) > 0 Graph rising at x (a) Although pictures of rising and falling graphs are suggestive, they provide no actual proof of the significance of the sign of the derivative. To establish rigorously the connection between a graph’s rising and falling and the sign of the derivative of the graphed function, we need the mean value theorem, stated later in this section. This theorem is the principal theoretical tool of differential calculus, and we shall see that it has many important applications. x f '(x) < 0 Graph falling at x (b) FIGURE 4.3.3 (a) A graph rising at x and (b) a graph falling at x. y Q P x FIGURE 4.3.4 Can you sail from P to Q without ever sailing—even for an instant—in the direction PQ (the direction of the arrow)? A Question As an introduction to the mean value theorem, we pose the following question. Suppose that P and Q are two points on the surface of the sea, with Q lying generally to the east of P (Fig. 4.3.4). Is it possible to sail a boat from P to Q, always sailing roughly east, without ever (even for an instant) sailing in the exact direction from P to Q? That is, can we sail from P to Q without our instantaneous line of motion ever being parallel to the line PQ? The mean value theorem answers this question: No. There will always be at least one instant when we are sailing parallel to the line PQ, no matter which path we choose. To paraphrase: Let the path of the sailboat be the graph of a differentiable function y = f (x) with endpoints P(a, f (a)) and Q(b, f (b)). Then we say that there must be some point on this graph where the tangent line (corresponding to the instantaneous line of motion of the boat) to the curve is parallel to the line PQ that joins the curve’s endpoints. This is a geometric interpretation of the mean value theorem. The Geometric Formulation The slope of the line tangent at the point (c, f (c)) (Fig. 4.3.5) is f (c), whereas the slope of the line PQ is f (b) − f (a) . b−a We may think of this last quotient as the average (or mean) value of the slope of the curve y = f (x) over the interval [a, b]. The mean value theorem guarantees that there is a point c in (a, b) for which the line tangent to y = f (x) at (c, f (c)) is indeed parallel to the line PQ. In the language of algebra, there’s a number c in (a, b) such that y Q(b, f (b)) y = f (x) f (c) = Slope f '(c) f (b) − f (a) . b−a (1) P(a, f(a)) c FIGURE 4.3.5 The sailboat problem in mathematical terminology. 236 x A Preliminary Result We first state a “lemma” to expedite the proof of the mean value theorem. This theorem is called Rolle’s theorem, after Michel Rolle (1652–1719), who discovered it in 1690. In his youth Rolle studied the emerging subject of calculus but later renounced it. He argued that the subject was based on logical fallacies, and he is remembered today only for the single theorem that bears his name. It is ironic that his theorem plays an important role in the rigorous proofs of several calculus theorems. Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 y f '(c) = 0 a c b Figure 4.3.6 illustrates the first case in the following proof of Rolle’s theorem. The idea of the proof is this: Suppose that the smooth graph y = f (x) starts (x = a) at height zero and ends (x = b) at height zero. Then if it goes up, it must come back down. But where it stops going up and starts coming back down, its tangent line must be horizontal. Therefore the derivative is zero at that point. Proof of Rolle's Theorem Because f is continuous on [a, b ], it must attain both a maximum and a minimum value on [a, b] (by the maximum value property of Section 3.5). If f has any positive values, consider its maximum value f (c). Now c is not an endpoint of [a, b] because f (a) = 0 and f (b) = 0. Therefore c is a point of (a, b). But we know that f is differentiable at c. So it follows from Theorem 2 of Section 3.5 that f (c) = 0. Similarly, if f has any negative values, we can consider its minimum value f (c) and conclude much as before that f (c) = 0. If f has neither positive nor negative values, then f is identically zero on [a, b ], and it follows that f (c) = 0 for every c in (a, b). Thus we see that the conclusion of Rolle’s theorem is justified in every case. ◆ x x x An important consequence of Rolle’s theorem is that between each pair of zeros of a differentiable function, there is at least one point at which the tangent line is horizontal. Some possible pictures of the situation are indicated in Fig. 4.3.7. FIGURE 4.3.7 The existence of the horizontal tangent is a consequence of Rolle’s theorem. y y = x1/2 − x3/2 EXAMPLE 2 Suppose that f (x) = x 1/2 − x 3/2 on [0, 1]. Find a number c that satisfies the conclusion of Rolle’s theorem. Solution Note that f is continuous on [0, 1] and differentiable on (0, 1). Because the term x 1/2 is present, f is not differentiable at x = 0, but this is irrelevant. Also, f (0) = 0 = f (1), so all of the hypotheses of Rolle’s theorem are satisfied. Finally, f (x) = 12 x −1/2 − 32 x 1/2 = 12 x −1/2 (1 − 3x), 0.2 0 so we see that f (c) = 0 for c = 13 . An accurate graph of f on [0, 1], including c and ◗ the horizontal tangent line, is shown in Fig. 4.3.8. c − 0.2 0 0.4 EXAMPLE 3 Suppose that f (x) = 1−x 2/3 on [−1, 1]. Then f satisfies the hypotheses of Rolle’s theorem except for the fact that f (0) does not exist. It is clear from the graph of f that there is no point where the tangent line is horizontal (Fig. 4.3.9). Indeed, 2 2 , f (x) = − x −1/3 = − √ 3 33x 0.8 x FIGURE 4.3.8 The number c of Example 2. so f (x) = 0 for x = 0, and we see that | f (x)| → ∞ as x → 0. Hence the graph of f has a vertical tangent line—rather than a horizontal one—at the point (0, 1). Thus the conclusion of Rolle’s theorem—like that of any theorem—may fail to hold if any ◗ of its hypotheses are not satisfied. 1.5 1 y = 1 − x2 /3 Now we are ready to state formally and prove the mean value theorem. y 0.5 0 −0.5 −1 ROLLE'S THEOREM Suppose that the function f is continuous on the closed interval [a, b] and is differentiable in its interior (a, b). If f (a) = 0 = f (b), then there exists some number c in (a, b) such that f (c) = 0. x FIGURE 4.3.6 The idea of the proof of Rolle’s theorem. 0.4 237 − 0.5 0 x 0.5 FIGURE 4.3.9 The function f (x) = 1 − x 2/3 of Example 3. 1 The Mean Value Theorem Suppose that the function f is continuous on the closed interval [a, b] and differentiable on the open interval (a, b). Then f (b) − f (a) = f (c) · (b − a) (2) for some number c in (a, b). 237 238 CHAPTER 4 Additional Applications of the Derivative COMMENT Because Eq. (2) is equivalent to Eq. (1), the conclusion of the mean value theorem is that there must be at least one point on the curve y = f (x) at which the tangent line is parallel to the line joining its endpoints P(a, f (a)) and Q(b, f (b)). y Q (b, f(b)) φ (x) P(a, f (a)) a x b x FIGURE 4.3.10 The construction of the auxiliary function φ. Motivation for the Proof of the Mean Value Theorem We consider the auxiliary function φ suggested by Fig. 4.3.10. The value of φ(x) is, by definition, the vertical height difference over x between the point (x, f (x)) on the curve and the corresponding point on the line PQ. It appears that a point on the curve y = f (x) where the tangent line is parallel to PQ corresponds to a maximum or minimum of φ. It’s also clear that φ(a) = 0 = φ(b), so Rolle’s theorem can be applied to the function φ on [a, b]. So our plan for proving the mean value theorem is this: First, we obtain a formula for the function φ. Second, we locate the point c such that φ (c) = 0. Finally, we show that this number c is exactly the number needed to satisfy the conclusion of the mean value theorem in Eq. (2). Proof of the Mean Value Theorem P(a, f (a)) and has slope Because the line PQ passes through f (b) − f (a) , b−a the point-slope formula for the equation of a straight line gives us the following equation for PQ: m= y = yline = f (a) + m(x − a). Thus φ(x) = ycurve − yline = f (x) − f (a) − m(x − a). You may verify by direct substitution that φ(a) = 0 = φ(b). And, because φ is continuous on [a, b] and differentiable on (a, b), we may apply Rolle’s theorem to it. Thus there is a point c somewhere in the open interval (a, b) at which φ (c) = 0. But φ (x) = f (x) − m = f (x) − f (b) − f (a) . b−a Because φ (c) = 0, we conclude that 0 = f (c) − f (b) − f (a) . b−a That is, f (b) − f (a) = f (c) · (b − a). ◆ The proof of the mean value theorem is an application of Rolle’s theorem, whereas Rolle’s theorem is the special case of the mean value theorem in which f (a) = 0 = f (b). 238 Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 239 EXAMPLE 4 Suppose that we drive from Kristiansand, Norway to Oslo—a road distance of almost exactly 350 km—in exactly 4 h, from time t = 0 to time t = 4. Let f (t) denote the distance we have traveled at time t, and assume that f is a differentiable function. Then the mean value theorem implies that 350 = f (4) − f (0) = f (c) · (4 − 0) = 4 f (c) and thus that f (c) = 350 4 = 87.5 at some instant c in (0, 4). But f (c) is our instantaneous velocity at time t = c, and 87.5 km/h is our average velocity for the trip. Thus the mean value theorem implies that we must have an instantaneous velocity of exactly 87.5 km/h at least once during ◗ the trip. The argument in Example 4 is quite general—during any trip, the instantaneous velocity must at some instant equal the average velocity for the whole trip. For instance, it follows that if two toll stations are 70 mi apart and you drive between the two in exactly 1 h, then at some instant you must have been speeding in excess of the posted limit of 65 mi/h. Speeding tickets have been issued by the Pennsylvania State Police to speeders on the Pennsylvania Turnpike on exactly such evidence! Consequences of the Mean Value Theorem The first of three important consequences of the mean value theorem is the nontrivial converse of the trivial fact that the derivative of a constant function is identically zero. That is, we prove that there can be no exotic function that is nonconstant but has a derivative that is identically zero. In Corollaries 1 through 3 we assume, as in Rolle’s theorem and the mean value theorem, that f and g are continuous on the closed interval [a, b] and differentiable on (a, b). COROLLARY 1 Functions with Zero Derivative If f (x) ≡ 0 on (a, b) (that is, f (x) = 0 for all x in (a, b)), then f is a constant function on [a, b]. In other words, there exists a constant C such that f (x) ≡ C. Proof Apply the mean value theorem to the function f on the interval [a, x], where x is a fixed but arbitrary point of the interval (a, b]. We find that f (x) − f (a) = f (c) · (x − a) for some number c between a and x. But f (x) is always zero on the interval (a, b), so f (c) = 0. Thus f (x) − f (a) = 0, and therefore f (x) = f (a). But this last equation holds for all x in (a, b]. Therefore, f (x) = f (a) for all x in (a, b] and, indeed, for all x in [a, b]. That is, f (x) has the constant value C = f (a). ◆ This establishes Corollary 1. Corollary 1 is usually applied in a different but equivalent form, which we state and prove next. COROLLARY 2 Functions with Equal Derivatives Suppose that f (x) = g (x) for all x in the open interval (a, b). Then f and g differ by a constant on [a, b]. That is, there exists a constant K such that f (x) = g(x) + K for all x in [a, b]. Proof Given the hypotheses, let h(x) = f (x) − g(x). Then h (x) = f (x) − g (x) = 0 239 240 CHAPTER 4 Additional Applications of the Derivative for all x in (a, b). So, by Corollary 1, h(x) is a constant K on [a, b ]. That is, f (x) − g(x) = K for all x in [a, b ]; therefore, f (x) = g(x) + K for all x in [a, b]. This establishes Corollary 2. ◆ EXAMPLE 5 If f (x) = 6e2x and f (0) = 7, what is the function f (x)? Solution Because Dx (e2x ) = 2e2x , we see immediately that one function with derivative g (x) = 6e2x is g(x) = 3e2x . Hence Corollary 2 implies that there exists a constant K such that f (x) = g(x) + K = 3e2x + K on any given interval [a, b] containing zero. But we can find the value of K by substituting x = 0: f (0) = 3e0 + K ; 7 = 3 · 1 + K; so K = 4. Thus the function f is defined by f (x) = 3e2x + 4. ◗ The following consequence of the mean value theorem verifies the remarks about increasing and decreasing functions with which we opened this section. COROLLARY 3 Increasing and Decreasing Functions If f (x) > 0 for all x in (a, b), then f is an increasing function on [a, b ]. If f (x) < 0 for all x in (a, b), then f is a decreasing function on [a, b ]. Proof Suppose, for example, that f (x) > 0 for all x in (a, b). We need to show the following: If u and v are points of [a, b] with u < v, then f (u) < f (v). We apply the mean value theorem to f , but on the closed interval [u, v]. This is legitimate because [u, v] is contained in [a, b], so f satisfies the hypotheses of the mean value theorem on [u, v] as well as on [a, b]. The result is that f (v) − f (u) = f (c) · (v − u) for some number c in (u, v). Because v > u and because, by hypothesis, f (c) > 0, it follows that f (v) − f (u) > 0; that is, f (u) < f (v), as we wanted to show. The proof is similar in the case that f (x) is negative on ◆ (a, b). The meaning of Corollary 3 is summarized in Fig. 4.3.11. Figure 4.3.12 shows a graph y = f (x) labeled in accord with this correspondence between the sign of the derivative f (x) and the increasing or decreasing behavior of the function f (x). EXAMPLE 6 Where is the function f (x) = x 2 − 4x + 5 increasing, and where is it decreasing? Solution The derivative of f is f (x) = 2x − 4. Clearly f (x) > 0 if x > 2, whereas f (x) < 0 if x < 2. Hence f is decreasing on (−∞, 2) and increasing on (2, +∞), as ◗ we see in Fig. 4.3.13. 240 Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 y f ' positive: f increasing 241 f ' negative: f decreasing x f (x) f (x) Negative Positive Decreasing Increasing f ' positive: f increasing FIGURE 4.3.12 The significance of the sign of f (x). FIGURE 4.3.11 Corollary 3. EXAMPLE 7 Show that the equation e x + x − 2 = 0 has exactly one [real] solution. 8 Solution A solution of the given equation will be a zero of the function y = x2 − 4x + 5 f (x) = e x + x − 2. y 4 Now f (0) = −1 < 0 while f (1) = e − 1 > 0. Because f is continuous (everywhere), the intermediate value property of continuous functions therefore guarantees that f (x) has at least one zero x0 in the interval (0, 1). We see this zero in Fig. 4.3.14, but cannot conclude from graphical evidence alone that there is no other zero somewhere (perhaps outside the viewing window of the figure). To prove that there is no other zero, we note that f is an increasing function on the whole real line. This follows from Corollary 3 and the fact that (2, 1) 0 0 2 x 4 FIGURE 4.3.13 The parabola of Example 6. f (x) = e x + 1 > 1 > 0 because e x > 0 for all x. Hence it follows from the definition of an increasing function that if x < x0 , then f (x) < f (x0 ) = 0, while if x > x0 then f (x) > f (x0 ) = 0. Thus x0 is the only zero of f (x) and hence is the one and only real solution of the equation e x + x − 2 = 0. ◗ 2 1.5 1 0.5 y ? 0 − 0.5 EXAMPLE 8 Determine the open intervals on the x-axis on which the function y = ex + x − 2 −1 f (x) = 3x 4 − 4x 3 − 12x 2 + 5 −1.5 −2 −1 −0.5 is increasing and those on which it is decreasing. 0 0.5 x 1 1.5 2 Solution The derivative of f is f (x) = 12x 3 − 12x 2 − 24x FIGURE 4.3.14 The graph y = e x + x − 2. = 12x(x 2 − x − 2) = 12x(x + 1)(x − 2). (3) The critical points x = −1, 0, and 2 separate the x-axis into the four open intervals (−∞, −1), (−1, 0), (0, 2), and (2, +∞) (Fig. 4.3.15). The derivative f (x) does not change sign within any of these intervals, because • • • The factor x + 1 in Eq. (3) changes sign only at x = −1, The factor 12x changes sign only at x = 0, and The factor x − 2 changes sign only at x = 2. x = −1 x+1<0 x−2<0 x=0 x+1>0 x−2<0 x=2 x+1>0 x−2>0 FIGURE 4.3.15 The signs of x + 1 and x − 2 (Example 8). 241 242 CHAPTER 4 Additional Applications of the Derivative Figure 4.3.15 indicates the signs of x + 1 and x − 2 on each of the four intervals. We illustrate two different methods of determining the sign of f (x) on each interval. Method 1 The second, third, and fourth columns of the next table record the signs of the factors in Eq. (3) on each of the four intervals listed in the first column. The signs of f (x) shown in the fifth column are then obtained by multiplication. The sixth column lists the resulting increasing or decreasing behavior of f on the four intervals. Interval x +1 12x x −2 f (x) (−∞, −1) (−1, 0) (0, 2) (2, +∞) − + + + − − + + − − − + − + − + f Decreasing Increasing Decreasing Increasing Method 2 Because the derivative f (x) does not change sign within any of the four intervals, we need only calculate its value at a single point in each interval. Whatever the sign at that point may be, it is the sign of f (x) throughout that interval. f (−2) = −96 < 0; f (−0.5) = 7.5 > 0; f (1) = −24 < 0; f (3) = 144 > 0; In (−∞, −1): In (−1, 0): In (0, 2): In (2, +∞): f f f f is decreasing. is increasing. is decreasing. is increasing. The second method is especially convenient if the derivative is complicated but an appropriate calculator for computation of its values is available. Finally, note that the results we have obtained in each method are consistent with ◗ the graph of y = f (x) shown in Fig. 4.3.16. y=x 40 1 20 y = sin x (0, 5) y 0 y 0 (−1, 0) −20 −1 (2, −27) −40 −1 0 1 −π 2 x FIGURE 4.3.16 The critical points of the polynomial of Example 8. 0 x π FIGURE 4.3.17 x and sin x (Example 9). EXAMPLE 9 The graph in Fig. 4.3.17 suggests that sin x < x for all x > 0. To show that this is indeed so, it suffices to show that the difference h(x) = f (x) − g(x) = x − sin x of the functions f (x) = x and g(x) = sin x is positive-valued for x > 0. But h (x) = 1 − cos x > 0 for all x in the interval (0, 2π ), where cos x < 1. Hence Corollary 3 implies that h is an increasing function on the closed interval [0, 2π ]. Because h(0) = 0, it therefore 242 Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 243 follows that h(x) > 0 if 0 < x 2π . But if x > 2π then certainly h(x) = x − sin x > 2π − sin x > 0 because | sin x| 1 for all x. Thus we have proved that x − sin x = h(x) > 0, and hence that x > sin x for all x > 0. (Can you tell why it follows from this that ◗ x < sin x for all x < 0?) 4.3 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. By definition, the function f is increasing on the interval I = (a, b) if f (x) > 0 for all x in I . 2. The mean value theorem implies that if the function f is differentiable on (a, b) and continuous at a and at b, then the line through (a, f (a)) and (b, f (b)) is parallel to some line tangent to the graph of f . 3. The mean value theorem states that f (b) − f (a) = f (c). b−a 4. If f (x) = 0 for all x in (a, b), then f (x) = 0 for all x in (a, b). 5. If f (x) = g (x) for all x in (a, b), then there is a constant C such that f (x) = g(x) + C for all x in (a, b). 6. One consequence of the mean value theorem is that if f (x) > 0 for all x in (a, b), then f is increasing on (a, b). 7. If f (x) = x 2 − 4x + 5, then f is increasing on (−∞, 2) and decreasing on (2, +∞). 8. If f (x) = 3x 4 −4x 3 −12x 2 +5, then f is increasing on (−1, 0) and on (2, +∞), decreasing on (−∞, −1) and on (0, 2). 9. If x > 0 then sin x < x. 10. If f (x) < 0 for all x in (a, b) then f is decreasing on (a, b). 4.3 CONCEPTS: QUESTIONS AND DISCUSSION 1. It’s often said that “what goes up must come down.” Can you translate this common saying into a mathematical statement? Does it follow from results in this section? 2. Suppose that f (x) > 0 for all x in the open interval (a, b). Why does it follow that there exists an inverse function g such that g( f (x)) = x for all x in (a, b)? What is the domain of definition of g? 3. Continuing Question 2, explain why it follows from results in this section that the function f (x) = e x has an inverse function (g(x) = ln x) that is defined for all x > 0. 4. Why does it not follow from results in this section that the function f (x) = sin x has an inverse function g such that g( f (x)) = x for all x? Determine a maximal closed interval I containing the origin such that there does exist a function g such that g( f (x)) = x for all x in I . Does your function g agree with the function sin−1 on your calculator? 5. Repeat Question 4, except with (a) f (x) = tan x; (b) f (x) = cos x. 243 244 CHAPTER 4 Additional Applications of the Derivative 4.3 PROBLEMS For the functions in Problems 1 through 6, first determine (as in Example 8) the open intervals on the x-axis on which each function is increasing and those where it is decreasing. Then use this information to match the function to its graph, one of the six shown in Fig. 4.3.18. 4 4 2 2 y 0 y 0 −2 −2 −4 −4 −4 −2 (a) 0 x 2 −4 4 4 4 2 2 y 0 y 0 −2 −2 0 x 2 4 −4 −2 (c) 0 x 2 4 2 2 y 0 y 0 −2 −2 0 x 2 0 x 2 4 28. f (x) = 5x [0, 5] −x 5/3 ; [−1, 1] 30. f (x) = 1 − (2 − x)2/3 ; [1, 3] [0, 1] [−1, 1] 33. f (x) = 3x + 6x − 5; [−2, 1] √ 34. f (x) = x − 1 ; [2, 5] 2 (f) −2 0 x 2 4 2. f (x) = x 2 − 2x − 1 3. f (x) = x 2 + 4x + 1 4. f (x) = 14 x 3 − 3x 5. f (x) = 13 x 3 − 12 x 2 − 2x + 1 6. f (x) = 2x − 16 x 2 − 19 x 3 In Problems 7 through 10, the derivative f (x) and the value f (0) are given. Use the method of Example 5 to find the function f (x). √ 7. f (x) = 4x; f (0) = 5 8. f (x) = 3 x ; f (0) = 4 1 9. f (x) = 2 ; f (1) = 1 10. f (x) = 6e−3x ; f (0) = 3 x In Problems 11 through 24, determine (as in Example 8) the open intervals on the x-axis on which the function is increasing as well as those on which it is decreasing. If you have a graphics calculator or computer, plot the graph y = f (x) to see whether it agrees with your results. 244 [0, π] 2/3 32. f (x) = x 3 ; 1. f (x) = 4 − x 2 12. f (x) = 4 − 5x 13. f (x) = 8 − 2x 2 14. f (x) = 4x 2 + 8x + 13 15. f (x) = 6x − 2x 2 16. f (x) = x 3 − 12x + 17 17. f (x) = x 4 − 2x 2 + 1 [−3, 3] 27. f (x) = 2 sin x cos x; 31. f (x) = xe ; FIGURE 4.3.18 Problems 1 through 6. 11. f (x) = 3x + 2 [0, 2] 4 In Problems 32 through 36, show that the given function f satisfies the hypotheses of the mean value theorem on the indicated interval, and find all numbers c in that interval that satisfy the conclusion of that theorem. −4 4 25. f (x) = x 2 − 2x; 26. f (x) = 9x − x ; −4 −2 22. f (x) = x 2 e−2x ln 2x for x > 0 23. f (x) = (x − 1)2 e−x 24. f (x) = x In Problems 25 through 28, show that the given function satisfies the hypotheses of Rolle’s theorem on the indicated interval [a, b ], and find all numbers x in (a, b) that satisfy the conclusion of that theorem. x 4 −4 21. f (x) = xe−x/2 29. f (x) = 1 − |x|; −2 (d) −4 √ 19. f (x) = 3x 4 + 4x 3 − 12x 2 20. f (x) = x x 2 + 1 In Problems 29 through 31, show that the given function f does not satisfy the conclusion of Rolle’s theorem on the indicated interval. Which of the hypotheses does it fail to satisfy? −4 4 Why?] x [Note: f (x) doesn’t change sign at x = −1. x +1 2 −4 −4 (e) −2 (b) 18. f (x) = 35. f (x) = (x − 1)2/3 ; [1, 2] 1 36. f (x) = x + ; [2, 3] x In Problems 37 through 40, show that the given function f satisfies neither the hypotheses nor the conclusion of the mean value theorem on the indicated interval. 37. f (x) = |x − 2|; [1, 4] 38. f (x) = 1 + |x − 1|; 39. f (x) = [[x]] 40. f (x) = 3x 2/3 [0, 3] (the greatest integer function); ; [−1, 1] [−1, 1] In Problems 41 through 44, show that the given equation has exactly one solution in the indicated interval. 41. x 5 + 2x − 3 = 0; 42. e −x = x − 1; 43. x ln x = 3; [0, 1] [1, 2] [2, 4] 44. sin x = 3x − 1; [−1, 1] 45. A car is driving along a rural road where the speed limit is 70 mi/h. At 3:00 P. M . its odometer (measuring distance traveled) reads 8075 mi. At 3:18 P. M . it reads 8100 mi. Prove that the driver violated the speed limit at some instant between 3:00 and 3:18 P. M . Increasing and Decreasing Functions and the Mean Value Theorem SECTION 4.3 46. Suppose that a car’s speedometer reads 50 mi/h at 3:25 P. M . and 65 mi/h at 3:35 P. M . Prove that at some instant in this 10-minute time interval the car’s acceleration was exactly 90 mi/h2 . 47. Points A and B along Interstate Highway 80 in Nebraska are 60 miles apart. Two cars both pass point A at 9:00 A . M . and both pass point B at 10:00 A . M . Show that at some instant between 9:00 and 10:00 A . M . the two cars have the same velocity. (Suggestion: Consider the difference h(t) = f (t) − g(t) between the position functions of the two cars.) 48. Show that the function f (x) = x 2/3 does not satisfy the hypotheses of the mean value theorem on [−1, 27] but that nevertheless there is a number c in (−1, 27) such that f (27) − f (−1) f (c) = . 27 − (−1) 49. Prove that the function f (x) = (1 + x) 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 3/2 − 3 x 2 (a) Show that g (0) = 12 > 0. (b) Sketch the graph of g near x = 0. Is g increasing on any open interval containing x = 0? [Answer: No.] 60. Suppose that f is increasing on every closed interval [a, b] provided that 2 a < b. Prove that f is increasing on the unbounded open interval (2, +∞). Note that the principle you discover was used implicitly in Example 6 of this section. Approximations Problems 61 through 64 illustrate the use of the mean value theorem to approximate numerical values of functions. 61. Use the method of Example 9 with f (x) = cos x and g(x) = 1 − 12 x 2 to show that cos x > 1 − 12 x 2 for all x > 0 (Fig. 4.3.19). −1 is increasing on (0, +∞). Explain carefully how you could conclude that (1 + x)3/2 > 1 + 32 x for all x > 0. Suppose that f is a constant function on the interval [a, b]. Prove that f must be a linear function (a function whose graph is a straight line). Suppose that f (x) is a polynomial of degree n − 1 on the interval [a, b ]. Prove that f (x) must be a polynomial of degree n on [a, b ]. Suppose that there are k different points of [a, b ] at which the differentiable function f vanishes (is zero). Prove that f must vanish on at least k − 1 points of [a, b ]. √ x on (a) Apply the mean value theorem to f (x) = [100, 101] to show that √ 1 101 = 10 + √ 2 c for some number c in (100, √ 101). (b) Show that if 100 < c < 101, then 10 < c < 10.5, and√use this fact to conclude from part (a) that 10.0475 < 101 < 10.0500. Prove that the equation x 7 + x 5 + x 3 + 1 = 0 has exactly one real solution. (a) Show that Dx tan2 x = Dx sec2 x on the open interval (−π/2, π/2). (b) Conclude that there exists a constant C such that tan2 x = sec2 x + C for all x in (−π/2, π/2). Then evaluate C. Explain why the mean value theorem does not apply to the function f (x) = |x| on the interval [−1, 2]. Suppose that the function f is differentiable on the interval [−1, 2] and that f (−1) = −1 and f (2) = 5. Prove that there is a point on the graph of f at which the tangent line is parallel to the line with the equation y = 2x. Let f (x) = x 4 − x 3 + 7x 2 + 3x − 11. Prove that the graph of f has at least one horizontal tangent line. Let the function g be defined as follows: ⎧ ⎨ x + x 2 sin 1 if x = 0, x g(x) = 2 ⎩ 0 if x = 0. 245 2 2 y=x y = g(x) 1 1 y = cos x y 0 y 0 −1 −1 −2 −4 y = sin x y = g(x) −2 0 x 2 4 FIGURE 4.3.19 cos x and g(x) = 1 − 12 x 2 (Problem 61). −2 −4 −2 0 x 2 4 FIGURE 4.3.20 x, sin x, and g(x) = x − 16 x 3 (Problem 62). 62. (a) Use the method of Example 9 and the result of Problem 61 to show that sin x > x − 16 x 3 for all x > 0 (Fig. 4.3.20). (b) Use the results of Example 9 and part (a) to calculate the sine of a 5◦ angle accurate to three decimal places. 63. (a) Use the results of Problem 62(a) to show that cos x < 1 − 12 x 2 + 1 4 x 24 for all x > 0. (b) Use the results of Problem 61 and part (a) to calculate the cosine of a 10◦ angle accurate to three decimal places. 64. Let pn (x) = 1 − x + x3 xn x2 − + · · · + (−1)n 2! 3! n! for each positive integer n. (a) Use the method of Example 9 to show that e−x > p1 (x) = 1 − x for all x > 0. (b) Use the result of part (a) to show that e−x < p2 (x) = 1 − x + 12 x 2 for all x > 0. (c) Use the result of part (b) to show that e−x > p3 (x) = 1 − x + 12 x 2 − 16 x 3 for all x > 0. (d) Continue one step at a time in like manner until you have shown that p7 (x) < e−x < p8 (x) for all x < 0. Finally, substitute x = 1 in this inequality to show that e ≈ 2.718 accurate to three decimal places. 245 246 CHAPTER 4 Additional Applications of the Derivative 4.4 THE FIRST DERIVATIVE TEST AND APPLICATIONS In Section 3.5 we discussed maximum and minimum values of a function defined on a closed and bounded interval [a, b]. Now we consider extreme values of functions defined on more general domains, including open or unbounded intervals as well as closed and bounded intervals. The distinction between absolute and local extrema is important here. Let c be a point of the domain D of the function f . Then recall from Section 3.5 that f (c) is the (absolute) maximum value of f (x) on D provided that f (c) f (x) for all x in D, whereas the value f (c) is a local maximum value of f (x) if it is the maximum value of f (x) on some open interval containing c. Similarly, f (c) is the (absolute) minimum value of f (x) on D provided that f (c) f (x) for all x in D; f (c) is a local minimum value of f (x) if it is the minimum value of f (x) on some open interval containing c. Thus a local maximum value is one that is as large as or greater than any nearby value of f (x), and a local minimum value is one that is as small as or less than any nearby value. Figure 4.4.1 shows a typical example of a function that has neither an absolute maximum nor an absolute minimum value. But each of the two local extrema pictured there is an (absolute) extreme value on a sufficiently small open interval. y Maximum Minimum x FIGURE 4.4.1 Local extrema are absolute extrema on sufficiently small intervals. REMARK Absolute extreme values are sometimes called global extreme values, and local extreme values are sometimes called relative extreme values. Theorem 2 of Section 3.5 tells us that any extremum of the differentiable function f on an open interval I must occur at a critical point where the derivative vanishes: f (x) = 0. But the mere fact that f (c) = 0 does not, by itself, imply that the critical value f (c) is an extreme value of f . Figures 4.4.2 through 4.4.5 illustrate different possibilities for the nature of f (c): whether it is a local or global maximum or minimum value, or neither. A Test for Local Extrema What we need is a way to test whether, at the critical point x = c, the value f (c) is actually a maximum or a minimum value of f (x), either local or global. Figure 4.4.6 shows how such a test might be developed. Suppose that the function f is continuous at c and that c is an interior point of the domain of f —that is, f is defined on some open interval that contains c. If f is decreasing immediately to the left of c and increasing immediately to the right, then f (c) should be a local minimum value of f (x). But if 8 (1, 4) 4 40 6 y = x3 − 3x2 − 9x 2 y = x2 + 3 y 0 (0, 3) 0 x 2 FIGURE 4.4.2 The graph of f (x) = x 2 + 3. The local minimum value f (0) = 3 is also the global minimum value of f (x). 246 y 0 −4 −4 −2 y = x3 + 2 y 0 −2 0 −2 (0, 2) (−1, 5) 4 y 8 4 2 y = 4 − (x − 1)2 −4 −2 0 x 2 (3, −27) − 40 4 FIGURE 4.4.3 The graph of f (x) = 4 − (x − 1)2 . The local maximum value f (1) = 4 is also the global maximum value of f (x). −4 0 4 x FIGURE 4.4.4 The graph of f (x) = x 3 − 3x 2 − 9x. The local minimum value f (3) = −27 clearly is not the global minimum value. Similarly, the local maximum value f (−1) = 5 is not the global maximum value. −8 −2 0 x 2 FIGURE 4.4.5 The graph of f (x) = x 3 + 2. The critical value f (0) = 2 is neither a global nor a local extreme value of f (x). The First Derivative Test and Applications SECTION 4.4 247 y y y f increasing f '(x) > 0 f decreasing f '(x) < 0 a f decreasing f '(x) < 0 f increasing f '(x) > 0 f increasing f '(x) > 0 f increasing f '(x) > 0 c b x Minimum a c b x Maximum a c b x Neither FIGURE 4.4.6 The first derivative test. f is increasing immediately to the left of c and decreasing immediately to its right, then f (c) should be a local maximum. If f is increasing on both sides or decreasing on both sides, then f (c) should be neither a maximum value nor a minimum value of f (x). Moreover, we know from Corollary 3 in Section 4.3 that the sign of the derivative f (x) determines where f (x) is decreasing and where it is increasing: • • f (x) is decreasing where f (x) < 0; f (x) is increasing where f (x) > 0. In the following test for local extrema, we say that L a To the left of c c • R To the b right of c FIGURE 4.4.7 Open intervals to the left and right of the point c. • f (x) < 0 to the left of c if f (x) < 0 on some interval (a, c) of numbers immediately to the left of c, and that f (x) > 0 to the right of c if f (x) > 0 on some interval (c, b) of numbers immediately to the right of c, and so forth. (See Fig. 4.4.7.) Theorem 1 tells us how to use the signs of f (x) to the left and right of the point c to determine whether f (x) has a local maximum or local minimum value at x = c. THEOREM 1 The First Derivative Test for Local Extrema Suppose that the function f is continuous on the interval I and also is differentiable there except possibly at the interior point c of I . 1. If f (x) < 0 to the left of c and f (x) > 0 to the right of c, then f (c) is a local minimum value of f (x) on I . 2. If f (x) > 0 to the left of c and f (x) < 0 to the right of c, then f (c) is a local maximum value of f (x) on I . 3. If f (x) > 0 both to the left of c and to the right of c, or if f (x) < 0 both to the left of c and to the right of c, then f (c) is neither a maximum nor a minimum value of f (x). COMMENT Thus f (c) is a local extremum if the first derivative f (x) changes sign as x increases through c, and the direction of this sign change determines whether f (c) is a local maximum or a local minimum. A good way to remember the first derivative test for local extrema is simply to visualize Fig. 4.4.6. Proof We will prove only part 1; the other two parts have similar proofs. Suppose that the hypotheses of Theorem 1 hold: that f is continuous on the interval I , that c is an interior point of I , and that f is differentiable on I except possibly at x = c. Then there exist two intervals (a, c) and (c, b), each wholly contained in I , such that f (x) < 0 on (a, c) and f (x) > 0 on (c, b). Suppose that x is in (a, b). Then there are three cases to consider. First, if x < c, then x is in (a, c) and f is decreasing on (a, c ], so f (x) > f (c). Second, if x > c, then x is in (c, b) and f is increasing on [c, b), so again f (x) > f (c). Finally, if 247 248 CHAPTER 4 Additional Applications of the Derivative f dec x = c, then f (x) = f (c). Thus, for each x in (a, b), f (x) f (c). Therefore, by ◆ definition, f (c) is a local minimum value of f (x). f inc The idea of this proof is illustrated in Fig. 4.4.8. Part (a) shows f decreasing to the left of c and increasing to the right, so there must be a local minimum at x = c. Part (b) shows f increasing to the left of c and decreasing to the right, so f (c) is a local maximum value of f (x). In part (c), the derivative has the same sign on each side of c, and so there can be no extremum of any sort at x = c. c (a) f inc REMARK Figures 4.4.9 through 4.4.13 illustrate cases in which Theorem 1 applies, where the interval I is the entire real number line R. In Fig. 4.4.9 through 4.4.11, the origin c = 0 is a critical point because f (0) = 0. In Figs. 4.4.12 and 4.4.13, c = 0 is a critical point because f (0) does not exist. f dec Classification of Critical Points c Suppose that we have found the critical points of a function. Then we can attempt to classify them—as local maxima, local minima, or neither—by applying the first derivative test at each point in turn. Example 1 illustrates a procedure that can be used. (b) f dec f dec EXAMPLE 1 Find and classify the critical points of the function f (x) = 2x 3 − 3x 2 − 36x + 7. c Solution The derivative is (c) f (x) = 6x 2 − 6x − 36 = 6(x + 2)(x − 3), FIGURE 4.4.8 The three cases in the first derivative test. 0.8 so the critical points [where f (x) = 0] are x = −2 and x = 3. These two points separate the x-axis into the three open intervals (−∞, −2), (−2, 3), and (3, +∞). y = x2 0.4 y y 0 0.8 0.8 0.4 0.4 0 − 0.4 − 0.4 − 0.8 − 0.8 −1 − 0.5 0 x 0.5 −1 1 FIGURE 4.4.9 f (x) = x 2 , f (x) = 2x, a local minimum at x = 0. y y = −x3 0 − 0.4 − 0.8 − 0.5 0 x 0.5 −1 1 FIGURE 4.4.10 f (x) = −x 2 , f (x) = −2x, a local maximum at x = 0. 0.8 2 /3 0.4 y = x 0.4 0 y − 0.4 − 0.5 y = x1/3 0 − 0.8 −0.5 0 x 0.5 FIGURE 4.4.12 f (x) = x 2/3 , f (x) = 23 x −1/3 , a local minimum at x = 0. 1 −1 0 x 0.5 FIGURE 4.4.11 f (x) = −x 3 , f (x) = −3x 2 , no extremum at x = 0. − 0.4 − 0.8 248 y y = −x2 0.8 −1 (1) −0.5 0 x 0.5 FIGURE 4.4.13 f (x) = x 1/3 , f (x) = 13 x −2/3 , no extremum at x = 0. 1 1 The First Derivative Test and Applications SECTION 4.4 249 The derivative f (x) cannot change sign within any of these intervals. One reason is that the factor x + 2 in Eq. (1) changes sign only at −2, whereas the factor x − 3 changes sign only at 3 (Fig. 4.4.14). As in Example 8 of Section 4.3, we illustrate here two methods of determining the signs of f (x) on the intervals (−∞, −2), (−2, 3), and (3, +∞). x = −2 x+2<0 x+2>0 x−3<0 x−3>0 x=3 FIGURE 4.4.14 The signs of x + 2 and x − 3 (Example 1). Method 1 The second and third columns of the following table record (from Fig. 4.4.14) the signs of the factors x + 2 and x − 3 in Eq. (1) on the three intervals listed in the first column. The signs of f (x) in the fourth column are then obtained by multiplication. Interval x +2 x −3 f (x) (−∞, −2) − − + (−2, 3) + − − (3, +∞) + + + Method 2 Because the derivative f (x) does not change sign within any of the three intervals, we need to calculate its value only at a single point in each interval: 80 In (−∞, −2): In (−2, 3): In (3, +∞): y = f(x) f '(x) > 0 40 y 0 f (−3) = 36 > 0; f (0) = −36 < 0; f (4) = 36 > 0; −2 f increasing x = −2 3 − 40 f is positive; f is negative; f is positive. f '(x) < 0 f decreasing f '(x) > 0 f increasing x=3 FIGURE 4.4.15 The three intervals of Example 1. −80 −4 −2 0 x 2 FIGURE 4.4.16 y = f (x) (Example 1). 4 Figure 4.4.15 summarizes our information about the signs of f (x). Because f (x) is positive to the left and negative to the right of the critical point x = −2, the first derivative test implies that f (−2) = 51 is a local maximum value. Because f (x) is negative to the left and positive to the right of x = 3, it follows that f (3) = −74 is a local minimum value. The graph of y = f (x) in Fig. 4.4.16 confirms this classification of the critical points x = −2 and x = 3. ◗ Open-Interval Maximum-Minimum Problems In Section 3.6 we discussed applied maximum-minimum problems in which the values of the dependent variable are given by a function defined on a closed and bounded interval. Sometimes, though, the function f describing the variable to be maximized (or minimized) is defined on an open interval (a, b), possibly an unbounded open interval such as (1, +∞) or (−∞, +∞), and we cannot “close” the interval by adjoining endpoints. Typically, the reason is that | f (x)| → +∞ as x approaches a or b. But if f has only a single critical point in (a, b), then the first derivative test can tell us that f (c) is the desired extreme value and can even determine whether it is a maximum or a minimum value of f (x). 249 250 CHAPTER 4 Additional Applications of the Derivative EXAMPLE 2 Figure 4.4.17 shows the graph of the function 1.5 ( e, 2e ) 1 0.5 2 ln x , x which is defined on the open interval (0, +∞). Because f (x) = y = 2 ln x x y 2 1 2 2 · − 2 · ln x = 2 (1 − ln x), x x x x there is a lone critical point at x = e. Note that f (x) = 0 −0.5 −1 −5 0 5 10 15 20 x • • FIGURE 4.4.17 The graph 2 ln x . y= x If x < e, then ln x < 1, so f (x) > 0 if x < e; If x > e, then ln x > 1, so f (x) < 0 if x > e. Therefore the first derivative test implies that f (e) = 2/e is a local maximum value of f . Indeed, because f is increasing if 0 < x < e and decreasing if x > e, it ◗ follows that 2/e is the absolute maximum value of f . EXAMPLE 3 Find the (absolute) minimum value of f (x) = x + 4 x for 0 < x < +∞. Solution The derivative is f (x) = 1 − y f '(x) < 0 x f'( )> 4 y=x+ x 0 (2) The roots of the equation x2 − 4 =0 x2 are x = −2 and x = 2. But x = −2 is not in the open interval (0, +∞), so we have only the critical point x = 2 to consider. We see immediately from Eq. (2) that f (x) = (2, 4) • x 2 2π r h Si d e • f (x) < 0 to the left of x = 2 (because x 2 < 4 there), and f (x) > 0 to the right of x = 2 (because x 2 > 4 there). Therefore, the first derivative test implies that f (2) = 4 is a local minimum value. We note also that f (x) → +∞ as either x → 0+ or as x → +∞. Hence the graph of f must resemble Fig. 4.4.18, and we see that f (2) = 4 is in fact the absolute minimum ◗ value of f (x) on the entire interval (0, +∞). FIGURE 4.4.18 The graph of the function of Example 3. EXAMPLE 4 We must make a cylindrical can with volume 125 in.3 (about 2 L) by cutting its top and bottom from squares of metal and forming its curved side by bending a rectangular sheet of metal to match its ends. What radius r and height h of the can will minimize the total amount of material required for the rectangle and the two squares? Solution We assume that the corners cut from the two squares, shown in Fig. 4.4.19, are wasted but that there is no other waste. As the figure shows, the area of the total amount of sheet metal required is r To p 4 x2 − 4 = . x2 x2 2r A = 8r 2 + 2πr h. h Bo t t o m The volume of the resulting can is then V = πr 2 h = 125, so h = 125/(πr 2 ). Hence A is given as a function of r by FIGURE 4.4.19 The parts to make the cylindrical can of Example 4. 250 125 250 = 8r 2 + , 0 < r < +∞. 2 πr r The domain of A is the unbounded open interval (0, +∞) because r can have any positive value, so A(r ) is defined for every number r in (0, +∞). But A(r ) → +∞ as A(r ) = 8r 2 + 2πr · The First Derivative Test and Applications SECTION 4.4 251 r → 0+ and as r → +∞. So we cannot use the closed-interval maximum-minimum method. But we can use the first derivative test. The derivative of A(r ) is 16 125 250 dA = 16r − 2 = 2 r 3 − . dr r r 8 (3) ; that is, Thus the only critical point in (0, +∞) is where r 3 = 125 8 r = 3 125 = 52 = 2.5. 8 We see immediately from Eq. (3) that A A'(r) < 0 A' ( r) d A/dr < 0 to the left of r = 52 , because r 3 < • d A/dr > 0 to the right, where r 3 > 125 8 there, and 125 . 8 Therefore, the first derivative test implies that a local minimum value of A(r ) on (0, +∞) is >0 A(r) = 8r 2 + 150 • 250 r A r 5 2 5 2 =8· 5 2 2 + 250 5 2 = 150. Considering that A(r ) → +∞ as r → 0+ and as r → +∞, we see that the graph of A(r ) on (0, +∞) looks like Fig. 4.4.20. This clinches the fact that A( 52 ) = 150 is the absolute minimum value of A(r ). Therefore, we minimize the amount of material required by making a can with radius r = 2.5 in. and height FIGURE 4.4.20 Graph of the function of Example 4. h= 20 125 = ≈ 6.37 2 π(2.5) π (in.). ◗ The total amount of material used is 150 in.2 EXAMPLE 5 Find the length of the longest rod that can be carried horizontally around the corner from a hall 2 m wide into one that is 4 m wide. 2 L2 Solution The desired length is the minimum length L = L 1 + L 2 of the rod being carried around the corner in Fig. 4.4.21. We see from the two similar triangles in the figure that 2 4 = sin θ and = cos θ, L1 L2 so L 1 = 4 csc θ and L 2 = 2 sec θ. θ 2 L1 4 4 θ FIGURE 4.4.21 Carrying a rod around a corner (Example 5). Therefore, the length L = L 1 + L 2 of the rod is given as a function of θ by L(θ ) = 4 csc θ + 2 sec θ y 3 − π 2 on the open interval (0, π/2). Note that L(θ ) → +∞ as either θ → 0+ or as θ → (π/2)− . (Why?) The derivative of L(θ ) is y = tan x 2 θ π 2 x dL = −4 csc θ cot θ + 2 sec θ tan θ dθ 2 sin θ 2 sin3 θ − 4 cos3 θ 4 cos θ = =− 2 + cos2 θ sin θ sin2 θ cos2 θ = (2 cos θ )(tan3 θ − 2) Hence d L/dθ = 0 exactly when FIGURE 4.4.22 y = tan x (Example 5). tan θ = sin2 θ √ 3 2, . so θ ≈ 0.90 (4) (rad). We now see from Eq. (4) and from the graph of the tangent function (Fig. 4.4.22) that 251 252 CHAPTER 4 Additional Applications of the Derivative • • d L/dθ < 0 to the left of θ ≈ 0.90, where tan θ < d L/dθ > 0 to the right, where tan3 θ > 2. √ 3 2, so tan3 θ < 2, and Hence the graph of L resembles Fig. 4.4.23. This means that the absolute minimum value of L—and therefore the maximum length of the rod in question—is about L L(0.90) = 4 csc(0.90) + 2 sec(0.90), dL > 0 dθ dL < 0 dθ approximately 8.32 m. ◗ The method we used in Examples 3 through 5 to establish absolute extrema illustrates the following global version of the first derivative test. 50 8.32 0.90 π 2 θ FIGURE 4.4.23 The graph of L(θ) (Example 5). THEOREM 2 The First Derivative Test for Global Extrema Suppose that f is defined on an open interval I , either bounded or unbounded, and that f is differentiable at each point of I except possibly at the single critical point c where f is continuous. 1. If f (x) < 0 for all x in I with x < c and f (x) > 0 for all x in I with x > c, then f (c) is the absolute minimum value of f (x) on I . 2. If f (x) > 0 for all x in I with x < c and f (x) < 0 for all x in I with x > c, then f (c) is the absolute maximum value of f (x) on I . The proof of this theorem is essentially the same as that of Theorem 1. REMARK When the function f (x) has only one critical point c in an open interval I , Theorem 2 may apply to tell us either that f (c) is the absolute minimum or that it is the absolute maximum of f (x) on I . But it is good practice to verify your conclusion by sketching the graph as we did in Examples 3 through 5. 4.4 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Suppose that the function f is continuous on the interval I and also is differentiable there except possibly at the interior point c of I . If f (x) < 0 to the left of c and f (x) > 0 to the right of c, then f (c) is a local minimum value of f (x) on I . 2. Suppose that the function f is continuous on the interval I and also is differentiable there except possibly at the interior point c of I . If f (x) > 0 both to the left of c and to the right of c, then f (c) is not an extremum of f . 3. If f (x) = 2x 3 − 3x 2 − 36x + 7, then f (x) cannot change sign on the interval (−2, 3). 4. If f (x) = 2x 3 − 3x 2 − 36x + 7, then f (3) is a local maximum value of f . 5. If f (x) = 2x 3 − 3x 2 − 36x + 7, then f has two critical points. 4 6. The absolute maximum value of f (x) = x + , 0 < x < +∞, is f (2) = 4. x 7. The longest rod that can be carried horizontally from a hall 4 meters wide around the corner into a perpendicular hall 2 meters wide is 4 + 2 = 6 meters. 8. Suppose that f is defined on the open interval I and is differentiable at each point of I except possibly at the critical point c, where f is continuous. If f (x) > 0 for all x in I with x < c and f (x) < 0 for all x in I with x > c, then f (c) is the absolute maximum value of f on I . 9. Suppose that f is defined on the open interval I and is differentiable at each point of I except possibly at the critical point c, where f is continuous. If f (x) < 0 for all x in I with x < c and f (x) > 0 for all x in I with x > c, then f (c) is the absolute minimum value of f on I . 252 The First Derivative Test and Applications SECTION 4.4 253 10. The first derivative test cannot be applied to the function g(x) = |x| on the interval [−1, 2]. 4.4 CONCEPTS: QUESTIONS AND DISCUSSION 1. Suppose that the function f is continuous on the whole real line R. Sketch a possible graph—if any—of f in each of the following three cases. (a) f has two critical points a and b but neither a local minimum nor a local maximum anywhere. Discuss separately the various possibilities: as to whether f is, or is not, differentiable at a and/or b. (b) f has three critical points but only a single local extremum. (c) f has three local minima but only a single local maximum. 2. Suppose that f is a cubic polynomial with positive leading coefficient. List the possibilities—with a typical graph of each—for the number and types of critical points of f . 3. Repeat Question 2 for a quartic (fourth-degree) polynomial. 4. Repeat Question 2 for a quintic (fifth-degree) polynomial. 5. Can you show that the function f must have an absolute minimum value if f is a polynomial of even degree with positive leading coefficient? 6. Can you show that the function f must have an absolute minimum value if lim f (x) = lim f (x) = +∞ x→a + x→b− and f is continuous on the open interval (a, b)? 4.4 PROBLEMS Apply the first derivative test to classify each of the critical points of the functions in Problems 1 through 16 (local or global, maximum or minimum, or not an extremum). If you have a graphics calculator or computer, plot y = f (x) to see whether the appearance of the graph corresponds to your classification of the critical points. 1. f (x) = x 2 − 4x + 5 2. f (x) = 6x − x 2 3. f (x) = x 3 − 3x 2 + 5 4. f (x) = x 3 − 3x + 5 5. f (x) = x 3 − 3x 2 + 3x + 5 6. f (x) = 2x 3 + 3x 2 − 36x + 17 10. f (x) = 3x 5 − 5x 3 12. f (x) = x 2 + 2 x 9. f (x) = x 4 − 2x 2 9 11. f (x) = x + x 13. f (x) = xe−2x 17. f (x) = sin2 x; (0, 3) 18. f (x) = cos2 x; 19. f (x) = sin x; (−3, 3) 20. f (x) = cos x; 4 21. f (x) = sin x − x cos x; (−5, 5) 22. f (x) = cos x + x sin x; (−5, 5) 2 (−3, 3) In Problems 27 through 50, which are applied maximumminimum problems, use the first derivative test to verify your answer. 28. A long rectangular sheet of metal is to be made into a rain gutter by turning up two sides at right angles to the remaining center strip (Fig. 4.4.24). The rectangular cross section of the gutter is to have area 18 in.2 Find the minimum possible width of the sheet. 14. f (x) = x 2 e−x/3 15. f (x) = (x + 4)2 e−x/5 1 − ln x 16. f (x) = for x > 0 x In Problems 17 through 26, find and classify the critical points of the given function in the indicated open interval. You may find it useful to construct a table of signs as in Example 1. 3 26. f (x) = x 3 e−x−x ; 27. Determine two real numbers with difference 20 and minimum possible product. 7. f (x) = 10 + 60x + 9x 2 − 2x 3 8. f (x) = 27 − x 3 ln x ; (0, 5) x2 ln(1 + x) 24. f (x) = ; (0, 5) 1+x 25. f (x) = e x sin x; (−3, 3) 23. f (x) = (−1, 3) (0, 4) A = 18 (in.2 ) FIGURE 4.4.24 The rectangular cross section of the gutter of Problem 28. 29. Find the point (x, y) on the line 2x + y = 3 that is closest to the point (3, 2). 253 254 CHAPTER 4 Additional Applications of the Derivative 30. You must construct a closed rectangular box with volume 576 in.3 and with its bottom twice as long as it is wide (Fig. 4.4.25). Find the dimensions of the box that will minimize its total surface area. 41. What point or points on the curve y = x 2 are nearest the point (0, 2)? [Suggestion: The square of a distance is minimized exactly when the distance itself is minimized.] 42. What is the length of the shortest line segment lying wholly in the first quadrant tangent to the graph of y = 1/x and with its endpoints on the coordinate axes? 43. A rectangle has area 64 cm2 . A straight line is to be drawn from one corner of the rectangle to the midpoint of one of the two more distant sides. What is the minimum possible length of such a line? 2x x FIGURE 4.4.25 The box of Problem 30. 31. Repeat Problem 30, but use an open-topped rectangular box with volume 972 in.3 32. An open-topped cylindrical pot is to have volume 125 in.3 What dimensions will minimize the total amount of material used in making this pot (Fig. 4.4.26)? Neglect the thickness of the material and possible wastage. 44. An oil can is to have volume 1000 in.3 and is to be shaped like a cylinder with a flat bottom but capped by a hemisphere (Fig. 4.4.27). Neglect the thickness of the material of the can, and find the dimensions that will minimize the total amount of material needed to construct it. 2 A = 2π r 2 2 h Atop = π r 2 Aside = 2π rh h r FIGURE 4.4.26 The cylinder of Problems 32, 33, 38, and 39. 33. An open-topped cylindrical pot is to have volume 250 cm3 (Fig. 4.4.26). The material for the bottom of the pot costs 4/ c/cm2 ; that for its curved side costs 2/ c/cm2 . What dimensions will minimize the total cost of this pot? 34. Find the point (x, y) on the parabola y = 4 − x 2 that is closest to the point (3, 4). [Suggestion: The cubic equation that you should obtain has a small integer as one of its roots. Suggestion: Minimize the square of the distance.] 35. Show that the rectangle with area 100 and minimum perimeter is a square. 36. Show that the rectangular solid with a square base, volume 1000, and minimum total surface area is a cube. 37. A box with a square base and an open top is to have volume 62.5 in.3 Neglect the thickness of the material used to make the box, and find the dimensions that will minimize the amount of material used. 38. You need a tin can in the shape of a right circular cylinder of volume 16π cm3 (Fig. 4.4.26). What radius r and height h would minimize its total surface area (including top and bottom)? 39. The metal used to make the top and bottom of a cylindrical can (Fig. 4.4.26) costs 4/ c/in.2 ; the metal used for the sides 2 costs 2/ c/in. The volume of the can must be exactly 100 in.3 . What dimensions of the can would minimize its total cost? 40. Each page of a book will contain 30 in.2 of print, and each page must have 2-in. margins at top and bottom and a 1-in. margin at each side. What is the minimum possible area of such a page? 254 4 4 r x FIGURE 4.4.27 The oil can of Problem 44. FIGURE 4.4.28 Carrying a rod around a corner (Problem 45). 45. Find the exact length L of the longest rod that can be carried horizontally around a corner from a corridor 2 m wide into one 4 m wide. Do this by minimizing the length of the rod in Fig. 4.4.28 by minimizing the square of that length as a function of x. 46. Find the length of the shortest ladder that will reach from the ground, over a wall 8 ft high, to the side of a building 1 ft behind the wall. That is, minimize the length L = L 1 + L 2 shown in Fig. 4.4.29. L2 y 8 1 L1 Wall x Ground FIGURE 4.4.29 The ladder of Problem 46. 47. A sphere with fixed radius a is inscribed in a pyramid with a square base so that the sphere touches the base of the pyramid and also each of its four sides. Show that the minimum possible volume of the pyramid is 8/π times the volume of the sphere. [Suggestion: Use the two right The First Derivative Test and Applications SECTION 4.4 255 triangles in Fig. 4.4.30 to show that the volume of the pyramid is V = V (y) = 4a 2 y 2 . 3(y − 2a) This can be done easily with the aid of the angle θ and without the formula for tan(θ/2). ] Don’t forget the domain of V (y). y θ a a θ FIGURE 4.4.30 Cross section through the centers of the sphere and pyramid of Problem 47. 48. Two noisy discothèques, one four times as noisy as the other, are located on opposite ends of a block 1000 ft long. What is the quietest point on the block between the two discos? The intensity of noise at a point away from its source is proportional to the noisiness and inversely proportional to the square of the distance from the source. 49. A floored tent with fixed volume V is to be shaped like a pyramid with a square base and congruent sides (Fig. 4.4.31). What height y and base edge 2x would minimize its total surface area (including its floor)? y z x FIGURE 4.4.31 The tent of Problem 49. 50. Suppose that the distance from the building to the wall in Problem 46 is a and that the height of the wall is b. Show that the minimal length of the ladder is 3/2 L min = a 2/3 + b2/3 . Problems 51 and 52 deal with square-based rectangular boxes. Such a box is said to be closed if it has both a square base and a top (as well as four vertical sides), open if it has a base but no top. (Problems 51 through 55 here are in a certain sense “dual” to Problems 56 through 60 in Section 3.6. Compare corresponding problems to make sure you see the difference; one will be a closed-interval maximum-minimum problem and the other an open-interval maximum-minimum problem.) 51. Show that, among all closed rectangular boxes with square bases and a given fixed volume, the one with minimal total surface area is a cube. 52. Show that, among all open rectangular boxes with square bases and a given fixed volume, the one with minimal total surface area has height equal to half the length of the edge of its base. Problems 53 through 55 deal with cans in the shape of right circular cylinders. Such a can is said to be closed if it has both a circular base and a top (as well as a curved side), open if it has a base but no top. 53. Show that, among all closed cylindrical cans with a given fixed volume, the one with minimal total surface area has height equal to the diameter of its base. 54. Show that, among all open cylindrical cans with a given fixed volume, the one with minimal total surface area has height equal to the radius of its base. 55. Suppose that the base and curved side of a pop-top soft drink can have the same thickness. But the top is three times as thick as the base to prevent ripping when the can is opened. Show that, among all such cans with a given fixed volume, the one requiring the least amount (volume) of material to make—including the triply thick top—has height twice the diameter of its base. Perhaps this is why soft drink cans look somewhat taller than vegetable or soup cans. 56. Suppose that you want to construct a closed rectangular box with a square base and fixed volume V . Each of the six faces of the box—the base, top, and four vertical sides—costs a cents per square inch, and gluing each of the 12 edges costs b cents per inch of edge length. What shape should this box be in order to minimize its total cost? [Suggestion: Show that the critical points of the cost function are roots of a certain quartic equation that you can solve using a computer algebra system. You may even be able to solve it with pencil and paper alone; begin by grouping the two terms of highest degree.] 4.4 INVESTIGATION: Constructing a Candy Box with Lid A candy maker wants to package jelly beans in boxes each having a fixed volume V . Each box is to be an open rectangular box with square base of edge length x. (See Fig. 4.4.32.) In addition, the box is to have a square lid with a two-inch rim. Thus the box-with-lid actually consists of two open rectangular boxes—the x-by-x-by-y box itself with height y 2 (in.) and the x-by-x-by-2 lid (which fits the box very snugly). Your job as the firm’s design engineer is to determine the dimensions x and y that will 255 256 CHAPTER 4 Additional Applications of the Derivative minimize the total cost of the two open boxes that comprise a single candy box with lid. Assume that the box-with-lid is to be made using an attractive foil-covered cardboard that costs $1 per square foot and that its volume is to be V = 400 + 50n cubic inches. (For your personal design problem, choose an integer n between 1 and 10.) 2 2 h y r x x FIGURE 4.4.32 The square-based candy box with lid. FIGURE 4.4.33 The cylindrical candy box with lid. Your next task is to design a cylindrical box-with-lid as indicated in Fig. 4.4.33. Now the box proper and its lid are both open circular cylinders, but everything else is the same as in the previous problem—two-inch rim, $1 per square foot foil-covered cardboard, and volume V = 400+50n. What are the dimensions of the box of minimal cost? Which is less expensive to manufacture—the optimal rectangular box-with-lid or the optimal box-with-lid in the shape of a cylinder? 4.5 SIMPLE CURVE SKETCHING We can construct a reasonably accurate graph of the polynomial function f (x) = an x n + an−1 x n−1 + · · · + a2 x 2 + a1 x + a0 (1) by assembling the following information. 1. The critical points of f —that is, the points on the graph where the tangent line is horizontal, so that f (x) = 0. 2. The increasing/decreasing behavior of f —that is, the intervals on which f is increasing and those on which it is decreasing. 3. The behavior of f “at” infinity—that is, the behavior of f as x → +∞ and as x → −∞. The same information often is the key to understanding the structure of a graph that has been plotted with a calculator or computer. Behavior at Infinity To carry out the task in item 3, we write f (x) in the form f (x) = x n an + an−1 a0 a1 + · · · + n−1 + n . x x x Thus we conclude that the behavior of f (x) as x → ±∞ is much the same as that of its leading term an x n , because all the terms that have powers of x in the denominator approach zero as x → ±∞. In particular, if an > 0, then lim f (x) = +∞, x→∞ meaning that f (x) increases without bound as x → +∞. Also +∞ if n is even; lim f (x) = x→−∞ −∞ if n is odd. 256 (2) (3) Simple Curve Sketching SECTION 4.5 257 If an < 0, simply reverse the signs on the right-hand sides in Eqs. (2) and (3). It follows that the graph of any (nonconstant) polynomial function exhibits one of the four “behaviors as x → ±∞” that are illustrated in Fig. 4.5.1. y y 20 20 10 10 −3 −2 −1 −10 1 2 4 x 3 −3 −20 −2 −1 −10 −20 −30 −40 y 3 4 x y 40 300 200 100 20 −1 2 (b) Southwest-southeast if n is even and an < 0 (a) Northwest-northeast if n is even and an > 0 −2 1 1 2 3 5 x 4 −4 −20 −40 −2 −100 −200 −300 2 4 x (d) Northwest-southeast if n is odd and an < 0 (c) Southwest-northeast if n is odd and an > 0 FIGURE 4.5.1 The behavior of polynomial graphs as x → ±∞. Critical Points Every polynomial, such as f (x) in Eq. (1), is differentiable everywhere. So the critical points of f (x) are the roots of the polynomial equation f (x) = 0—that is, solutions of nan x n−1 + (n − 1)an−1 x n−2 + · · · + 2a2 x + a1 = 0. (4) Sometimes we can find all (real) solutions of such an equation by factoring, but most often in practice we must resort to numerical methods aided by calculator or computer. Increasing/Decreasing Behavior x c1 c2 c3 ck − 1 ck FIGURE 4.5.2 The zeros of f (x) divide the x-axis into intervals on which f (x) does not change sign. Suppose that we have somehow found all the (real) solutions c1 , c2 , . . . , ck of Eq. (4). Then these solutions are the critical points of f . If they are arranged in increasing order, as in Fig. 4.5.2, then they separate the x-axis into the finite number of open intervals (−∞, c1 ), (c1 , c2 ), (c2 , c3 ), ... , (ck−1 , ck ), (ck , +∞) that also appear in the figure. The intermediate value property applied to f (x) tells us that f (x) can change sign only at the critical points of f , so f (x) has only one sign on each of these open intervals. It is typical for f (x) to be negative on some intervals and positive on others. Moreover, it’s easy to find the sign of f (x) on any one such interval I : We need only substitute any convenient number in I into f (x). Once we know the sign of f (x) on each of these intervals, we know where f is increasing and where it is decreasing. We then apply the first derivative test to find which of the critical values are local maxima, which are local minima, and which are neither—merely places where the tangent line is horizontal. With this information, the knowledge of the behavior of f as x → ±∞, and the fact that f is continuous, we can sketch its graph. We plot the critical points (ci , f (ci )) and connect them with a smooth curve that is consistent with our other data. 257 258 CHAPTER 4 Additional Applications of the Derivative It may also be helpful to plot the y-intercept (0, f (0)) and also any x-intercepts that are easy to find. But we recommend (until inflection points are introduced in Section 4.6) that you plot only these points—critical points and intercepts—and rely otherwise on the increasing and decreasing behavior of f . EXAMPLE 1 Sketch the graph of f (x) = x 3 − 27x. Solution Because the leading term is x 3 , we see that lim f (x) = +∞ and x→+∞ lim f (x) = −∞. x→−∞ Moreover, because f (x) = 3x 2 − 27 = 3(x + 3)(x − 3), (5) we see that the critical points where f (x) = 0 are x = −3 and x = 3. The corresponding points on the graph of f are (−3, 54) and (3, −54). The critical points separate the x-axis into the three open intervals (−∞, −3), (−3, 3), and (3, +∞) (Fig. 4.5.3). f '(x) > 0 x = −3 f (x) increasing f (−3) = 54 x=3 f '(x) < 0 f(x) decreasing f '(x) > 0 f(x) increasing f (3) = −54 FIGURE 4.5.3 The three open intervals of Example 1. To determine the increasing or decreasing behavior of f on these intervals, let’s substitute a number in each interval into the derivative in Eq. (5): On (−∞, −3): f (−4) = (3)(−1)(−7) = 21 > 0; f is increasing; On (−3, 3): f (0) = (3)(3)(−3) = −27 < 0; f is decreasing; On (3, +∞): f (4) = (3)(7)(1) = 21 > 0; f is increasing. √ √ We plot the critical points and the intercepts (0, 0), (3 3, 0), and (−3 3, 0). Then we use the information about where f is increasing or decreasing to connect these points with a smooth curve. Remembering that there are horizontal tangents at the two critical points, we obtain the graph shown in Fig. 4.5.4. y (−3, 54) Local maximum 50 25 + + + + + − − − − − − − − + + + + + −6 −5 −4 −3 −2 −1 (−3 3, 0) x-intercept 1 2 3 4 (0, 0) x-intercept −25 y-intercept 6 (3 3, 0) x-intercept y = x3 − 27x −50 (3, −54) Local minimum FIGURE 4.5.4 Graph of the function of Example 1. 258 5 x Simple Curve Sketching SECTION 4.5 259 In the figure we use plus and minus signs to mark the sign of f (x) in each interval. This makes it clear that (−3, 54) is a local maximum and that (3, −54) is a ◗ local minimum. The limits we found at the outset show that neither is global. EXAMPLE 2 Sketch the graph of f (x) = 8x 5 − 5x 4 − 20x 3 . Solution Because f (x) = 40x 4 − 20x 3 − 60x 2 = 20x 2 (x + 1)(2x − 3), (6) the critical points where f (x) = 0 are x = −1, x = 0, and x = 32 . These three critical points separate the x-axis into the four open intervals shown in Fig. 4.5.5. x = −1 x=0 x = 32 f '(x) > 0 f '(x) < 0 f '(x) < 0 f '(x) > 0 f(x) increasing f(x) decreasing f(x) decreasing f(x) increasing f(−1) = 7 f(0) = 0 f ( ) ≈ −32.06 3 2 FIGURE 4.5.5 The four open intervals of Example 2. This time, let’s determine the increasing or decreasing behavior of f by recording the signs of the factors in Eq. (6) on each of the subintervals shown in Fig. 4.5.5. In this way we get the following table: Interval x +1 20x 2 2x − 3 f (x) (−∞, −1) − + − + Increasing (−1, 0) 3 0, 2 3 , +∞ 2 + + − − Decreasing + + − − Decreasing + + + + Increasing f The points on the graph that correspond to the critical points are (−1, 7), (0, 0), and (1.5, −32.0625). We write f (x) in the form f (x) = x 3 (8x 2 − 5x − 20) in order to use the quadratic formula to find the x-intercepts. They turn out to be (−1.30, 0), (1.92, 0) (the abscissas are given only approximately), and the origin (0, 0). The latter is also the y-intercept. We apply the first derivative test to the increasing or decreasing behavior shown in the table. It follows that (−1, 7) is a local maximum, (1.5, −32.0625) is a local minimum, and (0, 0) is neither. The graph re◗ sembles the one shown in Fig. 4.5.6. In Example 3, the function is not a polynomial. Nevertheless, the methods of this section suffice for sketching its graph. EXAMPLE 3 Sketch the graph of f (x) = x 2/3 (x 2 − 2x − 6) = x 8/3 − 2x 5/3 − 6x 2/3 . Solution The derivative of f is f (x) = 83 x 5/3 − 10 2/3 x 3 − 12 −1/3 x 3 = 23 x −1/3 (4x 2 − 5x − 6) = 2(4x + 3)(x − 2) . 3x 1/3 (7) The tangent line is horizontal at the two critical points x = − 34 and x = 2, where the numerator in the last fraction of Eq. (7) is zero (and the denominator is not). Moreover, 259 260 CHAPTER 4 Additional Applications of the Derivative y y = 8x5 − 5x 4 − 20x 3 20 (−1, 7) Local maximum Horizontal tangent (0, 0) 10 (1.92, 0) ++++++++++ −2 +−−− −−−−− −−−−−−−− +++ −1 1 ++++++ x 2 (−1.30, 0) −10 −20 −30 (1.5, −32.06) Local minimum FIGURE 4.5.6 Graph of the function of Example 2. because of the presence of the factor x 1/3 in the denominator, | f (x)| → +∞ as x → 0. Thus x = 0 (a critical point because f is not differentiable there) is a point where the tangent line is vertical. These three critical points separate the x-axis into the four open intervals shown in Fig. 4.5.7. x = − 34 x=2 x=0 f '(x) < 0 f '(x) > 0 f '(x) < 0 f '(x) > 0 f (x) decreasing f(x) increasing f(x) decreasing f (x) increasing ( ) f − 34 ≈ −3.25 f (0) = 0 f (2) ≈ −9.52 FIGURE 4.5.7 The four open intervals of Example 3. We determine the increasing or decreasing behavior of f by substituting a number from each interval in f (x) (Eq. (7)). On −∞, − 34 : On − 34 , 0 : 2 · (−1)(−3) < 0; 3 · (−1) 1 2 · (+1) − 52 f −2 = 1/3 > 0; 3 · − 12 f (−1) = f is decreasing; f is increasing; On (0, 2) : f (1) = 2 · (+7)(−1) < 0; 3 · (+1) f is decreasing; On (2, +∞) : f (3) = 2 · (+15)(+1) > 0; 3 · (+3)1/3 f is increasing. The three critical points x = − 34 , x = 0, and x = 2 give the points (−0.75, −3.25), (0, 0), and (2, −9.52) on the graph (using approximations where appropriate). 260 Simple Curve Sketching SECTION 4.5 261 The first derivative test now shows local minima at (−0.75, −3.25) and at (2, −9.52); there is a local maximum at (0, 0). Although f (0) does not exist, the function f is continuous everywhere (because it involves only positive integral powers of x). We use the quadratic formula to find the x-intercepts. In addition to√the origin, 2 they occur √ where x − 2x − 6 = 0, and thus they are located at (1 − 7, 0) and at (1 + 7, 0). We then plot the approximations (−1.65, 0) and (3.65, 0). Finally, we note that f (x) → +∞ as x → ±∞. So the graph has the shape shown in Fig. 4.5.8. ◗ y 15 y = x2/3(x2 − 2x − 6) 10 (0, 0) Local maximum, vertical tangent 5 −2 −1 (−1.65, 0) (−0.75, −3.25) −5 Local minimum 1 2 x 4 (3.65, 0) 3 Local minimum (2, −9.52) −10 FIGURE 4.5.8 The technique is effective for nonpolynomial functions, as in Example 3. Curve Sketching and Solution of Equations An important application of curve-sketching techniques is the solution of an equation of the form f (x) = 0. (8) The real (as opposed to complex) solutions of this equation are simply the x-intercepts of the graph of y = f (x). Hence by sketching this graph with reasonable accuracy— either “by hand” or with a calculator or computer—we can glean information about the number of real solutions of Eq. (8) as well as their approximate locations. For example, Figs. 4.5.9 through 4.5.11 show the graphs of the cubic polynomials on the left-hand sides of the equations x 3 − 3x + 1 = 0, (9) x 3 − 3x + 2 = 0, (10) x − 3x + 3 = 0. (11) 3 Note that the polynomials differ only in their constant terms. 6 4 6 4 y = x3 − 3x + 1 6 y= x3 − 3x + 2 4 y 2 y 2 y 2 0 0 0 −2 −2 −2 −4 −2 0 x 2 4 FIGURE 4.5.9 y = x 3 − 3x + 1. −4 −2 y = x3 − 3x + 3 0 x 2 4 FIGURE 4.5.10 y = x 3 − 3x + 2. −4 −2 0 x 2 4 FIGURE 4.5.11 y = x 3 − 3x + 3. 261 262 CHAPTER 4 Additional Applications of the Derivative It is clear from Fig. 4.5.9 that Eq. (9) has three real solutions, one in each of the intervals [−2, −1], [0, 1], and [1, 2]. These solutions could be approximated graphically by successive magnification or analytically by Newton’s method. (As we have previously mentioned, there are even formulas—Cardan’s formulas—for the exact solution of an arbitrary cubic equation, but they are unwieldy and are seldom used except in computer algebra programs. For example, these formulas yield (via a computer algebra system) the expressions √ −1/3 √ 1/3 −1 + i 3 −1 + i 3 + , x1 = 2 2 √ 4/3 √ 5/3 −1 + i 3 −1 + i 3 + , (12) x2 = 2 2 and √ 2/3 √ 7/3 −1 + i 3 −1 + i 3 x3 = + 2 2 for the √ three solutions of Eq. (9). Despite the appearance of the imaginary number i = −1 in these three expressions, Fig. 4.5.9—with its three x-intercepts—indicates that all three solutions simplify to ordinary real numbers.) It appears in Fig. 4.5.10 that Eq. (10) has the two real solutions x = 1 and x = −2. Once we verify that x = 1 is a solution, then it follows from the factor theorem of algebra that x − 1 is a factor of x 3 − 3x + 2. The other factor can be found by division (long or synthetic) of x − 1 into x 3 − 3x + 2; the quotient is x 2 + x − 2. Thus we see that x 3 − 3x + 2 = (x − 1)(x 2 + x − 2) = (x − 1)2 (x + 2). Hence x = 1 is a “double root” and x = −2 is a “single root” of Eq. (10), thereby accounting for the three solutions that a cubic equation “ought to have.” We see in Fig. 4.5.11 that Eq. (11) has only one real solution. It is given approximately by x ≈ −2.1038. Problem 55 asks you to divide x + 2.0138 into x 3 − 3x + 3 to obtain a factorization of the form x 3 − 3x + 3 ≈ (x + 2.1038)(x 2 + bx + c). (13) The quadratic equation x 2 + bx + c = 0 has two complex conjugate solutions, which are the other two solutions of Eq. (12). Calculator and Computer Graphing 10 8 6 4 2 y 0 −2 −4 −6 −8 −10 −3 With a graphing calculator or computer we may construct the graph of a given function with a few keystrokes. Nevertheless, the viewpoint of this section may be useful in analyzing and understanding what we see on the screen. EXAMPLE 4 Figure 4.5.12 shows a computer-generated graph of the function f (x) = x 4 − 5x 2 + x + 2. −2 −1 0 x 1 2 FIGURE 4.5.12 y = x 4 − 5x 2 + x + 2. 262 3 (14) Three critical points are visible, separating the x-axis into two intervals on which the function f increases and two on which it decreases. In order to find these critical points, we need to solve the cubic equation f (x) = 4x 3 − 10x + 1 = 0. (15) For this purpose we could graph the derivative f (x) and zoom in on its solutions, or use Newton’s method to approximate these solutions accurately, or simply use the “solve” command on our calculator or computer. The approximate solutions of Eq. (15) thus found are −1.6289, 0.1004, and 1.5285. The corresponding numerical values of y obtained by substitution in Eq. (14) are −5.8554, 2.0501, and −2.6947. Thus the three critical points that we see on the graph in Fig. 4.5.12 are (−1.6289, −5.8554), (0.1004, 2.0501), and (1.5285, −2.6947). The function f is Simple Curve Sketching SECTION 4.5 263 decreasing on the intervals −∞ < x < −1.6289 and 0.1004 < x < 1.5285 and ◗ increasing on the intervals −1.6289 < x < 0.1004 and 1.55285 < x < ∞. 4.5 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If f (x) = x 3 − 27x, then lim f (x) = +∞. x→∞ 2. If f (x) = x − 27x, then lim f (x) = +∞. 3 x→−∞ 3. 4. 5. 6. 7. 8. 9. 10. If f (x) = x 3 − 27x, then f is increasing on the interval (−3, 3). If f (x) = 8x 5 − 5x 4 − 20x 3 , then f is decreasing on (−1, 0). If f (x) = 8x 5 − 5x 4 − 20x 3 , then f is decreasing on (0, 1.5). If f (x) = x 2/3 (x 2 − 2x − 6), then f is increasing on (0, 2). If f (x) = x 2/3 (x 2 − 2x − 6), then f has a local minimum at (0, 0). Every local maximum of the function f is also an absolute maximum of f . Cardan’s formulas are formulas for the solution of cubic equations. The equation x 3 − 3x + 2 = 0 has exactly two real solutions. 4.5 CONCEPTS: QUESTIONS AND DISCUSSION 1. Suppose that n is a positive integer and that k is an integer such that 0 k n. Does there always exist a polynomial of degree n having exactly k real zeros? If not, what are the exceptions? 2. Suppose that f (x) is a polynomial of degree n whose graph has p local minima and q local maxima. Explain why p + q < n. Discuss any other necessary restrictions on p and q. For instance, if n = 4 is it possible that p = 3 and q = 0? If n = 5 is it possible that p = 3 and q = 1? Justify your answers. 3. Someone asserts that “the graphs of any two n1th-degree polynomials with the same term of highest degree look essentially the same when plotted in a sufficiently large viewing window.” To what extent is this a reasonable claim? Begin by testing it with two quartic polynomials both having leading term x 4 . Do you need to adjust the x-scale, the y-scale, or both, to make the graphs nearly coincide? 4.5 PROBLEMS In Problems 1 through 4, use behavior “at infinity” to match the given function with its graph in Fig. 4.5.13. 1. f (x) = x 3 − 5x + 2 2. f (x) = x 4 − 3x 2 + x − 2 3. f (x) = − 13 x 5 − 3x 2 + 3x + 2 4. f (x) = − 13 x 6 + 2x 5 − 3x 4 + 12 x + 5 8 8 8 8 4 4 4 4 y 0 y 0 y 0 y 0 −4 −4 −4 −4 −8 −8 −4 (a) −2 0 x 2 −8 −8 −4 4 (b) −2 0 x 2 4 (c) −4 −2 0 x 2 −4 4 (d) −2 0 x 2 4 FIGURE 4.5.13 Problems 1 through 4. 263 264 CHAPTER 4 Additional Applications of the Derivative In Problems 5 through 14 a function y = f (x) and its computergenerated graph are given. Find both the critical points and the increasing/decreasing intervals for f (x). 5. y = 2x 2 − 10x − 7 (Fig. 4.5.14) 6. y = 27 + 12x − 4x 2 (Fig. 4.5.15) 50 40 30 20 10 y 0 −10 −20 −30 − 40 −50 −10 −5 0 x 5 50 40 30 20 10 y 0 −10 −20 −30 −40 −50 −8 −6 − 4 −2 0 2 4 6 8 x 10 14. y = 3x 8 − 52x 6 + 216x 4 − 500 (Fig. 4.5.23) × 104 1 0.8 0.6 0.4 0.2 y 0 −0.2 −0.4 −0.6 −0.8 −1 −6 −4 −2 0 x 2 4 2000 1500 1000 500 y 0 −500 −1000 −1500 −2000 −5 6 FIGURE 4.5.22 Problem 13. 0 x 5 FIGURE 4.5.23 Problem 14. 7. y = 4x − 3x − 90x + 23 (Fig. 4.5.16) In Problems 15 through 48, find the intervals on which the function f is increasing and those on which it is decreasing. Sketch the graph of y = f (x) and label the local maxima and minima. Global extrema should be so identified. 8. y = 85 + 70x − 11x 2 − 4x 3 (Fig. 4.5.17) 15. f (x) = 3x 2 − 6x + 5 FIGURE 4.5.14 Problem 5. 3 FIGURE 4.5.15 Problem 6. 2 250 200 150 100 50 y 0 −50 −100 −150 −200 −250 −8 −6 − 4 −2 0 2 4 6 8 x 250 200 150 100 50 y 0 −50 −100 −150 −200 −250 −8 −6 − 4 −2 0 2 4 6 8 x FIGURE 4.5.16 Problem 7. FIGURE 4.5.17 Problem 8. 200 150 100 50 y 0 −50 −100 −150 −200 −6 −4 −2 0 x 2 3 4 6 FIGURE 4.5.18 Problem 9. 4 800 600 400 200 y 0 −200 −400 −600 −800 −6 −4 −2 2 4 FIGURE 4.5.20 Problem 11. 21. f (x) = x 3 + 3x 2 + 9x 22. f (x) = x 3 − 27x 23. f (x) = (x − 1)2 (x + 2)2 32. f (x) = 6 − 5x − 6x 2 33. f (x) = 2x 3 + 3x 2 − 12x 0 x 2 4 6 FIGURE 4.5.19 Problem 10. 6 20. f (x) = x 3 + 6x 2 + 9x 29. f (x) = x 4 − 8x 2 + 7 1 30. f (x) = x 31. f (x) = 2x 2 − 3x − 9 3 0 x 19. f (x) = x 3 − 6x 2 + 9x 28. f (x) = x 4 + 4x 3 12. y = 2x 6 − 87x 4 + 600x 2 + 3000 (Fig. 4.5.21) 2000 1500 1000 500 y 0 −500 −1000 −1500 −2000 −6 − 4 −2 18. f (x) = x 3 + 3x 27. f (x) = 3x 5 − 5x 3 11. y = 3x − 100x + 960x (Fig. 4.5.20) 5 17. f (x) = x 3 − 12x 26. f (x) = x 2/3 (5 − x) 10. y = 125 + 120x − 2x − 9x (Fig. 4.5.19) 2 16. f (x) = 5 − 8x − 2x 2 24. f (x) = (x − 2)2 (2x + 3)2 √ √ 25. f (x) = 3 x − x x 9. y = 3x 4 + 4x 3 − 36x 2 + 40 (Fig. 4.5.18) 264 13. y = 3x 7 − 84x 5 + 448x 3 (Fig. 4.5.22) 8000 6000 4000 2000 y 0 −2000 −4000 −6000 −8000 −8 −6 −4 −2 0 2 4 6 8 x FIGURE 4.5.21 Problem 12. 34. f (x) = x 3 + 4x 35. f (x) = 50x 3 − 105x 2 + 72x 36. f (x) = x 3 − 3x 2 + 3x − 1 37. f (x) = 3x 4 − 4x 3 − 12x 2 + 8 38. f (x) = x 4 − 2x 2 + 1 39. f (x) = 3x 5 − 20x 3 40. f (x) = 3x 5 − 25x 3 + 60x 41. f (x) = 2x 3 + 3x 2 + 6x 42. f (x) = x 4 − 4x 3 43. f (x) = 8x 4 − x 8 44. f (x) = 1 − x 1/3 45. f (x) = x 1/3 (4 − x) 46. f (x) = x 2/3 (x 2 − 16) 47. f (x) = x(x − 1)2/3 48. f (x) = x 1/3 (2 − x)2/3 Simple Curve Sketching SECTION 4.5 265 In Problems 49 through 54, the values of the function f (x) at its critical points are given, together with the graph y = f (x) of its derivative. Use this information to construct a sketch of the graph y = f (x) of the function. 57. The computer-generated graph in Fig. 4.5.30 shows how the curve y = [x(x − 1)(2x − 1)]2 49. f (−3) = 78, f (2) = −47; Fig. 4.5.24 looks on any “reasonable” scale with integral units of measurement on the y-axis. Use the methods of this section to show that the graph really has the appearance shown in Fig. 4.5.31 (the values on the √ y-axis are in thousandths), with critical points at 0, 12 , 16 (3 ± 3 ), and 1. 50. f (−2) = 106, f (4) = −110; Fig. 4.5.25 60 60 40 40 y = f '(x) 20 y = f '(x) 20 15 0 6 −20 −20 4 −40 − 40 −60 −60 y 0 y −4 −2 0 x 2 4 FIGURE 4.5.24 y = f (x) of Problem 49. y −4 −2 0 x 2 4 FIGURE 4.5.25 y = f (x) of Problem 50. 51. f (−3) = −66, f (2) = 59; Fig. 4.5.26 60 100 40 y = f '(x) 20 y 0 y −20 0 −50 −40 −60 y = f '(x) 50 −100 −4 −2 0 x 2 −4 4 FIGURE 4.5.26 y = f (x) of Problem 51. −2 0 x 2 4 FIGURE 4.5.27 y = f (x) of Problem 52. 53. f (−2) = −107, f (1) = 82, f (3) = 18; Fig. 4.5.28 54. f (−3) = 5336, f (0) = 17, f (2) = 961, f (4) = −495; Fig. 4.5.29 100 2000 y = f '(x) 50 y 0 y −50 y = f '(x) −4 −2 0 x 2 0 −2000 −100 4 FIGURE 4.5.28 y = f (x) of Problem 53. −4 −2 0 x 2 10 2 y 0 4 FIGURE 4.5.29 y = f (x) of Problem 54. 55. (a) Verify the approximate solution x ≈ −2.0138 of Eq. (11). (b) Divide x 3 − 3x + 3 by x + 2.1038 to obtain the factorization in Eq. (13). (c) Use the quotient in part (b) to find (approximately) the complex conjugate pair of solutions of Eq. (11). 56. Explain why Figs. 4.5.9 and 4.5.10 imply that the cubic equation x 3 − 3x + q = 0 has exactly one real solution if |q| > 2 but has three distinct real solutions if |q| < 2. What is the situation if q = −2? 5 0 −2 −1 0 x 1 FIGURE 4.5.30 The graph y = [x(x − 1)(2x − 1)]2 on a “reasonable scale” (Problem 57). 52. f (−3) = −130, f (0) = 5, f (1) = −2; Fig. 4.5.27 × 10−3 2 −5 −0.5 0 0.5 x 1 1.5 FIGURE 4.5.31 The graph y = [x(x − 1)(2x − 1)]2 on a finer scale: −0.005 y 0.005 (Problem 57). 58. Use a computer algebra system to verify that the three expressions x1 , x2 , and x3 in Eq. (12) are, indeed, distinct real solutions of Eq. (9). Problems 59 and 60 require the use of a graphing calculator or computer algebra system. If you find it necessary to solve various equations, you may use either a graphing calculator or a “solve” command in a computer algebra system. 59. Show first that, on a “reasonable” scale with integral units of measurement on the y-axis, the graph of the polynomial 4 f (x) = 16 x(9x − 5)(x − 1) strongly resembles the graph shown in Fig. 4.5.30, with a seemingly flat section. Then produce a plot that reveals the true structure of the graph, as in Fig. 4.5.31. Finally, find the approximate coordinates of the local maximum and minimum points on the graph. 60. This problem pertains to the plausible suggestion that two polynomials with essentially the same coefficients ought to have essentially the same roots. (a) Show, nevertheless, that the quartic equation f (x) = x 4 − 55x 3 + 505x 2 + 11000x − 110000 = 0 has four distinct real solutions, whereas the “similar” equation g(x) = x 4 − 55x 3 + 506x 2 + 11000x − 110000 = 0 has only two distinct real solutions (and two complex conjugate solutions). (b) Let h(x) = f (x) + x 2 . Note that if = 0 then h(x) = f (x), and if = 1 then h(x) = g(x). Investigate the question of where—between = 0 and = 1—the transition from four real solutions of h(x) = 0 to only two real solutions takes place. 265 266 CHAPTER 4 Additional Applications of the Derivative 4.6 HIGHER DERIVATIVES AND CONCAVITY We saw in Section 4.3 that the sign of the first derivative f of a differentiable function f indicates whether the graph of f is rising or falling. Here we shall see that the sign of the second derivative of f , the derivative of f , indicates which way the curve y = f (x) is bending, upward or downward. Higher Derivatives The second derivative of f is the derivative of f ; it is denoted by f , and its value at x is f (x) = Dx ( f (x)) = Dx (Dx f (x)) = Dx2 f (x). (The superscript 2 is not an exponent but only an indication that the operator Dx is to be applied twice.) The derivative of f is the third derivative f of f , and f (x) = Dx ( f (x)) = Dx Dx2 f (x) = Dx3 f (x). The third derivative is also denoted by f (3) . More generally, the result of beginning with the function f and differentiating n times is succession is the nth derivative f (n) of f , with f (n) (x) = Dxn f (x). If y = f (x), then the first n derivatives are written in operator notation as Dx2 y, Dx y, Dx3 y, ... , Dxn y, y (x), ... , y (n) (x), in function notation as y (x), y (x), and in differential notation as dy d2 y d3 y dn y , , , ... , . 2 3 dx dx dx dxn The history of the curious use of superscripts in differential notation for higher derivatives involves the metamorphosis d dx dy dx → d2 y d dy (d)2 y → . → dx dx (d x)2 dx2 EXAMPLE 1 Find the first four derivatives of f (x) = 2x 3 + 1 + 16x 7/2 . x2 Solution Write f (x) = 2x 3 + x −2 + 16x 7/2 . Then 2 + 56x 5/2 , x3 6 f (x) = 12x + 6x −4 + 140x 3/2 = 12x + 4 + 140x 3/2 , x √ 24 f (x) = 12 − 24x −5 + 210x 1/2 = 12 − 5 + 210 x, x f (x) = 6x 2 − 2x −3 + 56x 5/2 = 6x 2 − and f (4) (x) = 120x −6 + 105x −1/2 = 266 120 105 +√ . x6 x ◗ Higher Derivatives and Concavity SECTION 4.6 267 Example 2 shows how to find higher derivatives of implicitly defined functions. EXAMPLE 2 Find the second derivative y (x) of a function y(x) that is defined implicitly by the equation x 2 − x y + y 2 = 9. Solution A first implicit differentiation of the given equation with respect to x gives 2x − y − x so dy dy + 2y = 0, dx dx dy y − 2x = . dx 2y − x We obtain d 2 y/d x 2 by differentiating implicitly, again with respect to x, using the quotient rule. After that, we substitute the expression we just found for dy/d x: y − 2x d y = Dx dx2 2y − x 2 dy dy − 2 (2y − x) − (y − 2x) 2 −1 dx dx (2y − x)2 = y − 2x dy − 3y 3x − 3y 2y − x dx = = . (2y − x)2 (2y − x)2 3x Thus 6(x 2 − x y + y 2 ) d2 y = − . dx2 (2y − x)3 We now substitute the original equation, x 2 − x y + y 2 = 9, for one final simplification: 54 d2 y =− . 2 dx (2y − x)3 The somewhat unexpected final simplification is always available when the original ◗ equation is symmetric in x and y. The Sign of the Second Derivative Now we shall investigate the significance of the sign of the second derivative. If f (x) > 0 on the interval I , then the first derivative f is an increasing function on I , because its derivative f (x) is positive. Thus, as we scan the graph y = f (x) from left to right, we see the tangent line turning counterclockwise (Fig. 4.6.1). We describe this situation by saying that the curve y = f (x) is bending upward. Note that a curve can bend upward without rising, as in Fig. 4.6.2. y y y = f(x) y = f (x) x FIGURE 4.6.1 The graph is bending upward. x FIGURE 4.6.2 Another graph bending upward. 267 268 CHAPTER 4 Additional Applications of the Derivative If f (x) < 0 on the interval I , then the first derivative f is decreasing on I , so the tangent line turns clockwise as x increases. We say in this case that the curve y = f (x) is bending downward. Figures 4.6.3 and 4.6.4 show two ways this can happen. The two cases are summarized in the brief table in Fig. 4.6.5. y y y = f(x) x f (x) y = f (x) Negative Positive Bending downward Bending upward y = f(x) x FIGURE 4.6.3 A graph bending downward. FIGURE 4.6.4 Another graph bending downward. EXAMPLE 3 Figure 4.6.6 shows the graph of the function y = x3 − 3x2 + 3 4 f (x) = x 3 − 3x 2 + 3. f"(x) < 0 2 (1, 1) Because y 0 f (x) = 3x 2 − 6x f"(x) > 0 −2 −2 0 x 2 and f (x) = 6x − 6 = 6(x − 1), we see that f (x) < 0 f (x) > 0 −4 −4 FIGURE 4.6.5 Significance of the sign of f (x) on an interval. 4 for for x < 1, x > 1. Observe in the figure that the curve bends downward on (−∞, 1) but bends upward on ◗ (1, +∞), consistent with the correspondences in Fig. 4.6.5. FIGURE 4.6.6 The graph of y = x 3 − 3x 2 + 3 (Example 3). The Second Derivative Test y y = f(x) = x3 x FIGURE 4.6.7 Although f (0) = 0, f (0) is not an extremum. 268 We know from Section 3.5 that a local extremum of a differentiable function f can occur only at a critical point where f (c) = 0, so the tangent line at the point (c, f (c)) on the curve y = f (x) is horizontal. But the example f (x) = x 3 , for which x = 0 is a critical point but not an extremum (Fig. 4.6.7), shows that the necessary condition f (c) = 0 is not a sufficient condition from which to conclude that f (c) is an extreme value of the function f . Now suppose not only that f (c) = 0, but also that the curve y = f (x) is bending upward on some open interval that contains the critical point x = c. It is apparent from Fig. 4.6.8(a) that f (c) is a local minimum value. Similarly, f (c) is a local maximum value if f (c) = 0 while y = f (x) is bending downward on some open interval containing c [Fig. 4.6.8(b)]. But the sign of the second derivative f (x) tells us whether y = f (x) is bending upward or downward and therefore provides us with a sufficient condition for a local extremum. THEOREM 1 Second Derivative Test Suppose that the function f is twice differentiable on the open interval I containing the critical point c at which f (c) = 0. Then 1. If f (x) > 0 on I , then f (c) is the minimum value of f (x) on I . 2. If f (x) < 0 on I , then f (c) is the maximum value of f (x) on I . Higher Derivatives and Concavity SECTION 4.6 269 y y y = f(x) (a) c y = f(x) x (b) c x FIGURE 4.6.8 The second derivative test (Theorem 1). (a) f (x) > 0: tangent turning counterclockwise; graph bending upward; local minimum at x = c. (b) f (x) < 0: tangent turning clockwise; graph bending downward; local maximum at x = c. f (x) f (c) Positive Negative Minimum Maximum FIGURE 4.6.9 Significance of the sign of f (x) on an interval containing the critical point c. y f(x) = x 4 x f "(0) = 0 — a local minimum y f "(0) = 0 — a local maximum x f(x) = −x 4 y f (x) = x 3 x f "(0) = 0 — neither a minimum nor a maximum FIGURE 4.6.10 No conclusion is possible if f (c) = 0 = f (c). Proof We will prove only part 1. If f (x) > 0 on I , then it follows that the first derivative f is an increasing function on I . Because f (c) = 0, we may conclude that f (x) < 0 for x < c in I and that f (x) > 0 for x > c in I . Consequently, the first derivative test of Section 4.4 implies that f (c) is the minimum value of ◆ f (x) on I . REMARK 1 Rather than memorizing verbatim the conditions in parts 1 and 2 of Theorem 1 (summarized in Fig. 4.6.9), it is easier and more reliable to remember the second derivative test by visualizing continuously turning tangent lines (Fig. 4.6.8). REMARK 2 Theorem 1 implies that the function f has a local minimum at the critical point c if f (x) > 0 on some open interval about c but a local maximum if f (x) < 0 near c. But the hypothesis on f (x) in Theorem 1 is global in that f (x) is assumed to have the same sign at every point of the open interval I that contains the critical point c. There is a strictly local version of the second derivative test that involves only the sign of f (c) at the critical point c (rather than on a whole open interval). According to Problem 90, if f (c) = 0, then f (c) is a local minimum value of f if f (c) > 0 but a local maximum if f (c) < 0. REMARK 3 The second derivative test says nothing about what happens if f (c) = 0 at the critical point c. Consider the three functions f (x) = x 4 , f (x) = −x 4 , and f (x) = x 3 . For each, f (0) = 0 and f (0) = 0. But their graphs, shown in Fig. 4.6.10, demonstrate that anything can happen at such a point—maximum, minimum, or neither. REMARK 4 Suppose that we want to maximize or minimize the function f on the open interval I , and we find that f has only one critical point in I , a number c at which f (c) = 0. If f (x) has the same sign at all points of I , then Theorem 1 implies that f (c) is an absolute extremum of f on I —a minimum if f (x) > 0 and a maximum if f (x) < 0. This absolute interpretation of the second derivative test can be useful in applied open-interval maximum-minimum problems. EXAMPLE 3 (continued) Consider again the function f (x) = x 3 − 3x 2 + 3, for which f (x) = 3x(x − 2) and f (x) = 6(x − 1). Then f has the two critical points x = 0 and x = 2, as marked in Fig. 4.6.6. Because f (x) < 0 for x near zero, the second derivative test implies that f (0) = 3 is a local maximum value of f . And because f (x) > 0 for x near 2, it follows that f (2) = −1 ◗ is a local minimum value. EXAMPLE 4 An open-topped rectangular box with square base has volume 500 cm3 . Find the dimensions that minimize the total area A of its base and four sides. 269 270 CHAPTER 4 Additional Applications of the Derivative Solution We denote by x the edge length of the square base and by y the height of the box (Fig. 4.6.11). The volume of the box is y x x FIGURE 4.6.11 The open-topped box of Example 4. V = x 2 y = 500, (1) and the total area of its base and four sides is A = x 2 + 4x y. (2) When we solve Eq. (1) for y = 500/x 2 and substitute this into Eq. (2), we get the area function A(x) = x 2 + 2000 , x 0 < x < +∞. The domain of A is the open and unbounded interval (0, +∞) because x can take on any positive value; to make the box volume 500, simply choose y = 500/x 2 . But x cannot be zero or negative. The first derivative of A(x) is A (x) = 2x − 2000 2(x 3 − 1000) = . x2 x2 (3) The equation A (x) = 0 yields x 3 = 1000, so the only critical point of A in (0, +∞) is x = 10. To investigate this critical point, we calculate the second derivative, A (x) = 2 + 4000 . x3 (4) Because it is clear that A (x) > 0 on (0, +∞), it follows from the second derivative test and Remark 4 that A(10) = 300 is the absolute minimum value of A(x) on (0, +∞). Finally, because y = 500/x 2 , y = 5 when x = 10. Therefore, this absolute ◗ minimum corresponds to a box with base 10 cm by 10 cm and height 5 cm. Concavity and Inflection Points A comparison of Fig. 4.6.1 with Fig. 4.6.3 suggests that the question of whether the curve y = f (x) is bending upward or downward is closely related to the question of whether it lies above or below its tangent lines. The latter question refers to the important property of concavity. DEFINITION Concavity Suppose that the function f is differentiable at the point a and that L is the line tangent to the graph y = f (x) at the point (a, f (a)). Then the function f (or its graph) is said to be 1. Concave upward at a if, on some open interval containing a, the graph of f lies above L. 2. Concave downward at a if, on some open interval containing a, the graph of f lies below L. Figure 4.6.12(a) shows a graph that is concave upward at (a, f (a)). Figure 4.6.12(b) shows a graph that is concave downward at (a, f (a)). Theorem 2 establishes the connection between concavity and the sign of the second derivative. That connection is the one suggested by our discussion of bending. THEOREM 2 Test for Concavity Suppose that the function f is twice differentiable on the open interval I . 1. If f (x) > 0 on I , then f is concave upward at each point of I . 2. If f (x) < 0 on I , then f is concave downward at each point of I . 270 Higher Derivatives and Concavity SECTION 4.6 271 y y = f (x) L (a, f(a)) x a (a) y (a, f(a)) L y = f(x) x a (b) FIGURE 4.6.12 (a) At x = a, f is a concave upward. (b) At x = a, f is concave downward. A proof of Theorem 2 based on the second derivative test is given at the end of this section. NOTE The significance of the sign of the first derivative must not be confused with the significance of the sign of the second derivative. The possibilities illustrated in Figs. 4.6.13 through 4.6.16 show that the signs of f and f are independent of each other. y y x x FIGURE 4.6.13 f (x) > 0, f increasing; f (x) > 0, f concave upward. FIGURE 4.6.14 f (x) > 0, f increasing; f (x) < 0; f concave downward. y y x x FIGURE 4.6.15 f (x) < 0, f decreasing; f (x) > 0, f concave upward. FIGURE 4.6.16 f (x) < 0, f decreasing; f (x) < 0, f concave downward. EXAMPLE 3 (continued again) For the function f (x) = x 3 − 3x 2 + 3, the second derivative changes sign from positive to negative at the point x = 1. Observe in Fig. 4.6.6 that the corresponding point (1, 1) on the graph of f is where the curve ◗ changes from bending downward to bending upward. 271 272 CHAPTER 4 Additional Applications of the Derivative Observe that the test for concavity in Theorem 2 says nothing about the case in which f (x) = 0. A point where the second derivative is zero may or may not be a point where the function changes from concave upward on one side to concave downward on the other. But a point like (1, 1) in Fig. 4.6.6, where the concavity does change in this manner, is called an inflection point of the graph of f . More precisely, the point x = a where f is continuous is an inflection point of the function f provided that f is concave upward on one side of x = a and concave downward on the other side. We also refer to (a, f (a)) as an inflection point on the graph of f . THEOREM 3 Inflection Point Test Suppose that the function f is continuous and f exists on an open interval containing the point a. Then a is an inflection point of f provided that f (x) < 0 on one side of a and f (x) > 0 on the other side. The fact that a point where the second derivative changes sign is an inflection point follows from Theorem 2 and the definition of an inflection point. REMARK At the inflection point itself, either • • f (a) = 0, or f (a) does not exist. Thus we find inflection points of f by examining the critical points of f . Some of the possibilities are indicated in Fig. 4.6.17. We mark the intervals of upward concavity and downward concavity by small cups opening upward and downward, respectively. y Vertical tangent Inflection point f"(a) = 0 and f'(a) = 0 Corner point that's also an inflection point f"(a) = 0, f'(a) ≠ 0 x FIGURE 4.6.17 Some inflection points. EXAMPLE 5 Figure 4.6.18 shows the graph of f (x) = (2x 2 − 3x − 1)e−x . Two evident inflection points are marked. Find their coordinates. 1 (9/2, 26e−9/2) 0.5 Solution We calculate 0 y= y −0.5 (2x2 − 3x − 1)e−x f (x) = (4x − 3)e−x − (2x 2 − 3x − 1)e−x = (−2x 2 + 7x − 2)e−x (1, −2e−1) −1 and −1.5 −2 −1 f (x) = (−4x + 7)e−x − (−2x 2 + 7x − 2)e−x = (2x 2 − 11x + 9)e−x . 0 1 2 3 x 4 FIGURE 4.6.18 The graph y = (2x 2 − 3x − 1)e−x (Example 5). 272 5 6 7 Because e−x is never zero, it follows that f (x) = 0 only when 2x 2 − 11x + 9 = (2x − 9)(x − 1) = 0 —that is, when either x = 1 or x = 92 . Only at these two points can f (x) change Higher Derivatives and Concavity SECTION 4.6 273 sign. But f (0) = 9 > 0, f (2) = −5e−x < 0, and f (5) = 4e−5 > 0. It therefore follows that f (x) > 0 if x < 1, f (x) < 0 if 1 < x < 92 , f (x) > 0 if 9 2 and < x. Thus the graph of f (x) = (2x 2 − 3x − 1)e−x has inflection points where x = 1 and where x = 92 . These points, marked on the graph in Fig. 4.6.18, have coordinates (1, −2e−1 ) and ( 92 , 26e−9/2 ). ◗ Inflection Points and Curve Sketching Let the function f be twice differentiable for all x. Just as the critical points where f (x) = 0 separate the x-axis into open intervals on which f (x) does not change sign, the possible inflection points where f (x) = 0 separate the x-axis into open intervals on which f (x) does not change sign. On each of these intervals, the curve y = f (x) either is bending downward [ f (x) < 0] or is bending upward [ f (x) > 0]. We can determine the sign of f (x) in each of these intervals in either of two ways: 1. Evaluation of f (x) at a typical point of each interval. The sign of f (x) at that particular point is the sign of f (x) throughout the interval. 2. Construction of a table of signs of the factors of f (x). Then the sign of f (x) on each interval can be deduced from the table. These are the same two methods we used in Sections 4.4 and 4.5 to determine the sign of f (x). We use the first method in Example 6 and the second in Example 7. EXAMPLE 6 Sketch the graph of f (x) = 8x 5 −5x 4 −20x 3 , indicating local extrema, inflection points, and concave structure. Solution We sketched this curve in Example 2 of Section 4.5; see Fig. 4.5.6 for the graph. In that example we found the first derivative to be f (x) = 40x 4 − 20x 3 − 60x 2 = 20x 2 (x + 1)(2x − 3), so the critical points are x = −1, x = 0, and x = 32 . The second derivative is f (x) = 160x 3 − 60x 2 − 120x = 20x(8x 2 − 3x − 6). When we compute f (x) at each critical point, we find that f (−1) = −100 < 0, f (0) = 0, and f 3 2 = 225 > 0. Continuity of f ensures that f (x) < 0 near the critical point x = −1 and that f (x) > 0 near the critical point x = 32 . The second derivative test therefore tells us that f has a local maximum at x = −1 and a local minimum at x = 32 . We cannot determine from the second derivative test the behavior of f at x = 0. Because f (x) exists everywhere, the possible inflection points are the solutions of the equation f (x) = 0; that is, 20x(8x 2 − 3x − 6) = 0. 273 274 CHAPTER 4 Additional Applications of the Derivative Clearly, one solution is x = 0. To find the other two, we use the quadratic formula to solve the equation 8x 2 − 3x − 6 = 0. This gives x= 1 16 √ 3 ± 201 , so x ≈ 1.07 and x ≈ −0.70 are possible inflection points along with x = 0. These three possible inflection points separate the x-axis into the intervals indicated in Fig. 4.6.19. We check the sign of f (x) on each. On (−∞, −0.70) : On (−0.70, 0) : On (0, 1.07) : On (1.07, +∞) : f (−1) f − 12 f (1) f (2) = = = = −100 < 0; 25 > 0; −20 < 0; 800 > 0; x = −0.70 x=0 f f f f is concave downward; is concave upward; is concave downward; is concave upward. x = 1.07 f"(x) < 0 f"(x) > 0 f"(x) < 0 f "(x) > 0 Bending down Bending up Bending down Bending up FIGURE 4.6.19 Intervals of concavity of Example 6. Thus we see that the direction of concavity of f changes at each of the three points x ≈ −0.70, x = 0, and x ≈ 1.07. These three points are indeed inflection ◗ points. This information is shown in the graph of f sketched in Fig 4.6.20. y y = 8x5 − 5x 4 − 20x3 20 Local maximum (−1, 7) 10 Horizontal tangent, inflection point (0, 0) (1.92, 0) −2 (−1.3, 0) −1 1 2 x Inflection point (−0.70, 4.30) −10 −20 −30 Inflection point (1.07, −19.98) Local minimum (1.5, −32.06) FIGURE 4.6.20 The graph of the function of Example 6. EXAMPLE 7 Sketch the graph of f (x) = 4x 1/3 + x 4/3 . Indicate local extrema, inflection points, and concave structure. 274 Higher Derivatives and Concavity SECTION 4.6 275 Solution First, f (x) = 4 −2/3 4 1/3 4(x + 1) + x = , x 3 3 3x 2/3 so the critical points are x = −1 (where the tangent line is horizontal) and x = 0 (where it is vertical). Next, 8 4 4(x − 2) , f (x) = − x −5/3 + x −2/3 = 9 9 9x 5/3 so the possible inflection points are x = 2 (where f (x) = 0) and x = 0 (where f (x) does not exist). To determine where f is increasing and where it is decreasing, we construct the following table. Interval x +1 x 2/3 f (x) (−∞, −1) (−1, 0) (0, +∞) − + + + + + − + + f Decreasing Increasing Increasing Thus f is decreasing when x < −1 and increasing when x > −1 (Fig. 4.6.21(a)). x = −1 x=0 f '(x) < 0 f '(x) > 0 f '(x) > 0 f decreasing f increasing f increasing (a) x=0 x=2 f "(x) > 0 f "(x) < 0 f "(x) > 0 Bending up Bending down Bending up (b) FIGURE 4.6.21 (a) Increasing and decreasing intervals of Example 7. (b) Intervals of concavity of Example 7. To determine the concavity of f , we construct a table to find the sign of f (x) on each of the intervals separated by its zeros. Interval x 5/3 x −2 f (x) (−∞, 0) (0, 2) (2, +∞) − + + − − + + − + f Concave upward Concave downward Concave upward The table shows that f is concave downward on (0, 2) and concave upward for x < 0 and for x > 2 (Fig. 4.6.21(b)). We note that f (x) → +∞ as x → ±∞, and we mark with plus signs the intervals on the x-axis where f is increasing, minus signs where it is decreasing, cups opening upward where f is concave upward, and cups opening downward where f is concave downward. We plot (at least approximately) the points on the graph of f that correspond to the zeros and discontinuities of f and f ; these are (−1, −3), (0, 0), √ 3 and (2, 6 2 ). Finally, we use all this information to draw the smooth curve shown in ◗ Fig. 4.6.22. 275 276 CHAPTER 4 Additional Applications of the Derivative y y = 4x1/3 + x 4/3 (2, 7.6) Inflection point − − − − − + + (−1, −3) Local minimum (0, 0) + + + + Vertical tangent x- and y-intercept inflection point + x FIGURE 4.6.22 The graph of the function of Example 7. EXAMPLE 8 The graph of the equation x 2 − x y + y 2 = 9 is the rotated ellipse shown in Fig. 4.6.23. In Example 2 we saw that if the function y(x) is implicitly defined by this equation, then y 4 x2 − xy + y2 3, 2 3 =9 2 −4 ( −2 y= ) x 2 2 dy y − 2x = . dx 2y − x 4 x −2 (− 3 , −2 3 ) −4 FIGURE 4.6.23 The ellipse x 2 − x y + y 2 = 9 is concave downward at points above the line y = 12 x, concave upward at points beneath it. y y = f(x) g(x) = f(x) − T(x) Hence y = 2x at any critical point (x, y) at which y (x) = 0. Substituting √ √y = 2x 2 2 − x y + y = 9 readily gives the two points ( 3 , 2 3 ) and into√the equation x √ (− 3 , −2 3 ) that are marked in Fig. 4.6.23. We also saw that 54 d2 y =− . 2 dx (2y − x)3 It follows that y (x) < 0 when 2y − x > 0, so the graph is concave downward at any point (x, y) at which 2y > x; that is, at points above the line y = 12 x. Similarly, y (x) > 0 when 2y − x < 0, so the graph is concave upward at any point (x, y) at ◗ which 2y < x; that is, at points below the line y = 12 x. (See Fig. 4.6.23.) Proof of Theorem 2 We will prove only part 1—the proof of part 2 is similar. Given a fixed point a of the open interval I where f (x) > 0, we want to show that the graph y = f (x) lies above the tangent line at (a, f (a)). The tangent line in question has the equation y = T (x) = f (a) + f (a) · (x − a). (5) Consider the auxiliary function g(x) = f (x) − T (x) (a, f (a)) (6) y = T(x) a x FIGURE 4.6.24 Illustrating the proof of Theorem 2. x illustrated in Fig. 4.6.24. Note first that g(a) = g (a) = 0, so x = a is a critical point of g. Moreover, Eq. (5) implies that T (x) ≡ f (a) and that T (x) ≡ 0, so g (x) = f (x) − T (x) = f (x) > 0 at each point of I . Therefore, the second derivative test implies that g(a) = 0 is the minimum value of g(x) = f (x) − T (x) on I . It follows that the curve y = f (x) lies ◆ above the tangent line y = T (x). 4.6 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. The second derivative of the function f is Dx ( f (x)). 1 120 105 2. If f (x) = 2x 3 + 2 + 16x 7/2 , then f (4) (x) = 6 + √ . x x x 276 Higher Derivatives and Concavity SECTION 4.6 277 d2 y 54 = . dx2 (2y − x)3 If f (x) > 0 on (a, b), then the graph of f is bending downward on (a, b). Suppose that the function f is twice differentiable on the open interval I containing the critical point c at which f (c) = 0. If f (x) < 0 on I , then f (c) is the maximum value of f (x) on I . If f (x) > 0 on (a, b), then the graph of f is concave upward on (a, b). Suppose that the function f is continuous and that f exists on an open interval containing the point a. Then a is an inflection point of f provided that f (x) < 0 on one side of a and f (x) > 0 on the other side. The graph of f (x) = 8x 5 − 5x 4 − 20x 3 has exactly three inflection points. The graph of f (x) = 4x 1/3 − x 4/3 has both a vertical tangent and an inflection point at (0, 0). The graph of f (x) = 4x 1/3 − x 4/3 is concave downward on (0, 2). 3. If x 2 − x y + y 2 = 9, then 4. 5. 6. 7. 8. 9. 10. 4.6 CONCEPTS: QUESTIONS AND DISCUSSION 1. Suppose that the function f is differentiable at the point x = c where f (c) = 0. Does it necessarily follow that the point (c, f (c)) is an inflection point of the graph of y = f (x)? 2. Suppose that the function f is differentiable except at the point x = c where the graph of y = f (x) has a vertical tangent line. Does it necessarily follow that the point (c, f (c)) is an inflection point of the graph of y = f (x)? 3. Suppose that n is a positive integer and that k is an integer such that 0 k n − 2. Does there always exist a polynomial of degree n having exactly k inflection points? Justify your answer. 4. Can the graph of a function have more inflection points than critical points? Justify your answer. 4.6 PROBLEMS Calculate the first three derivatives of the functions given in Problems 1 through 15. 20. 1. f (x) = 2x 4 − 3x 3 + 6x − 17 1 2. f (x) = 2x 5 + x 3/2 − 2x 2 3. f (x) = (2x − 1)2 √ 4. g(t) = t 2 + t + 1 √ 3 8. f (x) = 1 + x 9. g(t) = t 2 ln t 12. f (x) = cos2 2x 14. f (x) = x cos x 2 y y+1 24. The graph of f (x) = 2x 3 − 9x 2 − 108x + 200 (Fig. 4.6.26) 200 400 100 11. f (x) = sin 3x y 13. f (x) = sin x cos x sin x 15. f (x) = x In Problems 16 through 22, calculate dy/d x and d 2 y/d x 2 , assuming that y is defined implicitly as a function of x by the given equation. 16. x 2 + y 2 = 4 21. sin y = x y 23. The graph of f (x) = x 3 − 3x 2 − 45x (Fig. 4.6.25) 7. h(y) = z 1 1 + =1 x y In Problems 23 through 30, find the exact coordinates of the inflection points and critical points marked on the given graph. 5. g(t) = (3t − 2) √ 6. f (x) = x x + 1 e 10. h(z) = √ 19. y 3 + x 2 + x = 5 22. sin2 x + cos2 y = 1 4/3 z 18. x 1/3 + y 1/3 = 1 17. x 2 + x y + y 2 = 3 200 0 y −200 −100 −200 −10 0 − 400 −5 0 x 5 10 FIGURE 4.6.25 The graph of f (x) = x 3 − 3x 2 − 45x (Problem 23). −10 −5 0 x 5 10 FIGURE 4.6.26 The graph of f (x) = 2x 3 − 9x 2 − 108x + 200 (Problem 24). 277 278 CHAPTER 4 Additional Applications of the Derivative 25. The graph of f (x) = 4x 3 −6x 2 −189x +137 (Fig. 4.6.27) 26. The graph of f (x) = −40x 3 − 171x 2 + 2550x + 4150 (Fig. 4.6.28) × 104 y 800 1 400 0.5 0 −10 44. f (x) = sec x on (−π/2, π/2) 45. f (x) = cos2 x on (−π/2, 3π/2) 46. f (x) = sin3 x on (−π, π) −0.5 −800 −1 −5 0 x 5 −10 10 FIGURE 4.6.27 The graph of f (x) = 4x 3 − 6x 2 − 189x + 137 (Problem 25). −5 400 5 47. f (x) = 10(x − 1)e−2x 10 (Fig. 4.6.29) (Fig. 4.6.30) y − 400 0 x 5 0 −5 10 FIGURE 4.6.29 The graph of f (x) = x 4 − 54x 2 − 237 (Problem 27). 0 5 x 10 15 FIGURE 4.6.30 The graph of f (x) = x 4 − 10x 3 − 250 (Problem 28). 29. The graph of f (x) = 3x − 20x + 1000 (Fig. 4.6.31) 30. The graph of f (x) = 3x 5 − 160x 3 (Fig. 4.6.32) 5 1 −10 −1 −5 0 x 5 10 FIGURE 4.6.31 The graph of f (x) = 3x 5 − 20x 4 + 1000 (Problem 29). 35. f (x) = xe−x 37. f (x) = x 5 + 2x 39. f (x) = x 2 (x − 1)2 41. f (x) = sin x on (0, 2π ) 278 53. Problem 29 54. Problem 30 55. Problem 31 56. Problem 32 57. Problem 33 58. Problem 36 59. Problem 37 60. Problem 38 61. Problem 39 62. Problem 40 −2 −10 64. f (x) = 3x 4 − 4x 3 − 5 −5 0 x 5 10 FIGURE 4.6.32 The graph of f (x) = 3x 5 − 160x 3 (Problem 30). Apply the second derivative test to find the local maxima and local minima of the functions given in Problems 31 through 50, and apply the inflection point test to find all inflection points. 31. f (x) = x 2 − 4x + 3 33. f (x) = x 3 − 3x + 1 52. Problem 28 63. f (x) = 2x 3 − 3x 2 − 12x + 3 y 0 −2000 51. Problem 27 Sketch the graphs of the functions in Problems 63 through 76, indicating all critical points and inflection points. Apply the second derivative test at each critical point. Show the correct concave structure and indicate the behavior of f (x) as x → ±∞. × 104 2000 0 2 4 2 y 49. f (x) = (x 2 − 2x − 1)e−x In Problems 51 through 62, rework the indicated problem from Section 4.4, now using the second derivative test to verify that you have found the desired absolute maximum or minimum value. −1000 −5 48. f (x) = (x 2 − x)e−x 50. f (x) = xe−x 1000 0 −10 0 x FIGURE 4.6.28 The graph of f (x) = −40x 3 − 171x 2 + 2550x + 4150 (Problem 26). 27. The graph of f (x) = x 4 − 54x 2 + 237 28. The graph of f (x) = x 4 − 10x 3 − 250 y 43. f (x) = tan x on (−π/2, π/2) 0 y − 400 42. f (x) = cos x on (−π/2, π/2) 32. f (x) = 5 − 6x − x 2 34. f (x) = x 3 − 3x 2 ln x 36. f (x) = x 38. f (x) = x 4 − 8x 2 40. f (x) = x 3 (x + 2)2 65. f (x) = 6 + 8x 2 − x 4 66. f (x) = 3x 5 − 5x 3 67. f (x) = 3x 4 − 4x 3 − 12x 2 − 1 68. f (x) = 3x 5 − 25x 3 + 60x 69. f (x) = x 3 (x − 1)4 70. f (x) = (x − 1)2 (x + 2)3 71. f (x) = 1 + x 1/3 72. f (x) = 2 − (x − 3)1/3 √ 73. f (x) = (x + 3) x 74. f (x) = x 2/3 (5 − 2x) √ 75. f (x) = (4 − x) 3 x 76. f (x) = x 1/3 (6 − x)2/3 Higher Derivatives and Concavity SECTION 4.6 279 In Problems 77 through 82, the graph of a function f (x) is shown. Match it with the graph of its second derivative f (x) in Fig. 4.6.33. 8 8 4 4 y 0 y 0 −4 −4 −8 −8 −10 −5 0 x (a) 5 10 81. See Fig. 4.6.38. 82. See Fig. 4.6.39. 8 8 4 4 y 0 y 0 −4 −4 −8 −8 −10 −10 (b) −5 0 x 5 10 −5 0 x 5 10 FIGURE 4.6.38 −10 −5 0 x 5 10 FIGURE 4.6.39 83. (a) Show first that the nth derivative of f (x) = x n is 8 8 4 4 y 0 y 0 −4 −4 −8 −8 −10 −5 0 x (c) 5 10 f (n) (x) ≡ n! = n · (n − 1) · (n − 2) · · · 3 · 2 · 1. −10 (d) 8 8 4 4 y 0 y 0 −4 −4 −5 0 x 5 10 d2z d2z = dx2 dy 2 −8 −8 −10 −5 0 x (e) 5 −10 10 −5 0 x (f) 5 10 FIGURE 4.6.33 77. See Fig. 4.6.34. 78. See Fig. 4.6.35. 8 8 4 4 y 0 y 0 −4 −4 −8 −8 −10 −5 0 x 5 10 −10 −5 0 x 5 FIGURE 4.6.35 79. See Fig. 4.6.36. 80. See Fig. 4.6.37. 8 4 4 y 0 y 0 −4 −4 −8 −8 −10 −5 0 x FIGURE 4.6.36 5 10 −10 10 dy dx 2 + dz d 2 y . · dy d x 2 86. Prove that the graph of a quadratic polynomial has no inflection points. 87. Prove that the graph of a cubic polynomial has exactly one inflection point. 88. Prove that the graph of a polynomial function of degree 4 has either no inflection point or exactly two inflection points. 89. Suppose that the pressure p (in atmospheres), volume V (in cubic centimeters), and temperature T (in kelvins) of n moles of carbon dioxide (CO2 ) satisfy van der Waals’ equation p+ FIGURE 4.6.34 8 (b) Conclude that if f (x) is a polynomial of degree n, then f (k) (x) ≡ 0 if k > n. 84. (a) Calculate the first four derivatives of f (x) = sin x. (b) Explain why it follows that Dxn+4 sin x = Dxn sin x if n is a positive integer. 85. Suppose that z = g(y) and that y = f (x). Show that n2a (V − nb) = n RT, V2 where a, b, and R are empirical constants. The following experiment was carried out to find the values of these constants. One mole of CO2 was compressed at the constant temperature T = 304 K. The measured pressure-volume ( pV ) data were then plotted as in Fig. 4.6.40, with the pV curve showing an inflection point coinciding with a horizontal tangent at V = 128.1, p = 72.8. Use this information to calculate a, b, and R. [Suggestion: Solve van der Waals’ equation for p and then calculate dp/d V and d 2 p/d V 2 .] p (128.1, 72.8) V −5 0 x FIGURE 4.6.37 5 10 FIGURE 4.6.40 A problem involving van der Waals’ equation. 279 280 CHAPTER 4 Additional Applications of the Derivative 90. Suppose that the function f is differentiable on an open interval containing the point c at which f (c) = 0 and that the second derivative f (c) = lim h→0 f (c + h) − f (c) f (c + h) = lim h→0 h h exists. (a) First assume that f (c) > 0. Reason that f (c+h) and h have the same sign if h = 0 is sufficiently small. Hence apply the first derivative test to show in this case that f (c) is a local minimum value of f . (b) Show similarly that f (c) is a local maximum value of f if f (c) < 0. Problems 91 and 92 require the use of a graphing calculator or computer algebra system. Any equations you need to solve may be solved graphically or by using a “solve” key or command. 91. Figure 4.6.41 shows the graph of the cubic equation y = 1000x 3 − 3051x 2 + 3102x + 1050 on a scale with x measured in units and y in tens of thousands. The graph appears to exhibit a single point near 1 0.8 0.6 0.4 0.2 y 0 −0.2 −0.4 −0.6 −0.8 −1 × 10 4 (1, 2000) that is both a critical point and an inflection point. Nevertheless, this curve “has two real wiggles like a good cubic should.” Find them! In particular, find the local extrema and inflection point (or points) on this curve. Then sketch a graph that plainly exhibits all these points—mark and label each of them. 92. Figure 4.6.42 shows the graph of y = [x(1 − x)(9x − 7)(4x − 1)]4 on a scale with x and y both measured in units. At first glance it appears that there is a local maximum near x = 12 , with “flat spots” along the x-axis to the left and to the right. But no nonconstant polynomial can have a “flat spot” where y = 0 on an open interval of the x-axis. (Why not?) Indeed, this graph actually has seven local extrema and six inflection points in the interval 0 x 1. Find approximate coordinates of all thirteen of these points, then sketch a graph on a scale that makes all these points evident. 4 3 2 y 1 0 −1 −1 0 1 x 2 3 FIGURE 4.6.41 The cubic graph of Problem 91. −2 −0.5 0 0.5 x 1 1.5 FIGURE 4.6.42 The graph of Problem 92. 4.7 CURVE SKETCHING AND ASYMPTOTES We now extend the limit concept to include infinite limits and limits at infinity. This extension will add a powerful weapon to our arsenal of curve-sketching techniques, the notion of an asymptote to a curve—a straight line that the curve approaches arbitrarily close in a sense we soon make precise. Recall from Section 2.3 that f (x) is said to increase without bound, or become infinite, as x approaches a, and we write lim f (x) = +∞, x→a (1) provided that f (x) can be made arbitrarily large by choosing x sufficiently close (but not equal) to a. The statement that f (x) decreases without bound, or becomes negatively infinite, as x → a, written lim f (x) = −∞, x→a has an analogous definition. 280 (2) Curve Sketching and Asymptotes SECTION 4.7 281 EXAMPLE 1 It is apparent that 1 = +∞ x→−2 (x + 2)2 lim because, as x → −2, (x + 2)2 is positive and approaches zero. (See Fig. 4.7.1.) By contrast, x = −∞ x→−2 (x + 2)2 lim because, as x → −2, the denominator (x + 2)2 is still positive and approaches zero, but the numerator x is negative. (See Fig. 4.7.2.) Thus when x is very close to −2, we have in x/(x + 2)2 a negative number close to −2 divided by a very small positive ◗ number. Hence the quotient becomes a negative number of large magnitude. 15 y= 5 1 (x + 2)2 x = −2 10 0 y 5 y −5 y= x (x + 2)2 −10 0 x = −2 −5 −6 −4 −2 0 2 −15 −6 4 x y= x = −2 −2 0 2 4 x x → −∞ (x + 2)2 as x → −2, and the line x = −2 is a vertical asymptote. 1 → ∞ as (x + 2)2 x → −2, and the line x = −2 is a vertical asymptote. FIGURE 4.7.2 FIGURE 4.7.1 10 8 6 4 2 y 0 −2 −4 −6 −8 −10 −6 −4 One-sided versions of Eqs. (1) and (2) are valid also. For instance, if n is an odd positive integer, then it is apparent that 1 (x + 2) 3 lim x→−2− 1 = −∞ (x + 2)n and lim x→−2+ 1 = +∞, (x + 2)n because (x + 2)n is negative when x is to the left of −2 and positive when x is to the right of −2. The case n = 3 is illustrated in Fig. 4.7.3. Vertical Asymptotes −4 −2 0 2 4 x 1 has (x + 2)3 infinite one-sided limits as x → −2, and the line x = −2 is a vertical asymptote. FIGURE 4.7.3 The vertical lines at x = −2 in Figs. 4.7.1 through 4.7.3 are examples of vertical asymptotes associated with infinite limits. The line x = a is a vertical asymptote of the curve y = f (x) provided that either lim f (x) = ±∞ (3a) lim f (x) = ±∞ (3b) x→a − or x→a + or both. It is usually the case that both one-sided limits, rather than only one, are infinite. If so, we write lim f (x) = ±∞. x→a (3c) 281 282 CHAPTER 4 Additional Applications of the Derivative y x x=1 FIGURE 4.7.4 A ”right-hand only” vertical asymptote. y The geometry of a vertical asymptote is illustrated by the graphs in Figs. 4.7.1 through 4.7.3. In each case, as x → −2 and f (x) → ±∞, the point (x, f (x)) on the curve approaches the vertical asymptote x = −2 and the shape and direction of the curve are better and better approximated by the asymptote. Figure 4.7.4 shows the graph of a function whose left-hand limit is zero at x = 1. But the right-hand limit there is +∞, which explains why the line x = 1 is a vertical asymptote for this graph. The right-hand limit in Fig. 4.7.5 does not even exist, but because the left-hand limit at x = 1 is −∞, the vertical line at x = 1 is again a vertical asymptote. A vertical asymptote typically appears in the case of a rational function f (x) = p(x)/q(x) at a point x = a where q(a) = 0 but p(a) = 0. (See Examples 4 through 8 later in this section.) Limits at Infinity In Section 4.5 we discussed infinite limits at infinity in connection with the behavior of a polynomial as x → ±∞. There is also such a thing as a finite limit at infinity. We say that f (x) approaches the number L as x increases without bound and write x=1 lim f (x) = L (4) x→+∞ x provided that |f (x) − L| can be made arbitrarily small (close to zero) merely by choosing x sufficiently large. That is, given > 0, there exists M > 0 such that x>M implies | f (x) − L| < . (5) The statement that FIGURE 4.7.5 The behavior of the graph to its left produces the vertical asymptote. lim f (x) = L x→−∞ has a definition of similar form—merely replace the condition x > M with the condition x < −M. The analogues for limits at infinity of the limit laws of Section 2.2 all hold, including, in particular, the sum, product, and quotient laws. In addition, it is not difficult to show that if lim f (x) = L and x→+∞ lim g(x) = ±∞, x→+∞ then f (x) = 0. g(x) lim x→+∞ It follows from this result that lim x→+∞ 1 =0 xk (6) for any choice of the positive rational number k. Using Eq. (6) and the limit laws, we can easily evaluate limits at infinity of rational functions. The general method is this: First divide each term in both the numerator and the denominator by the highest power of x that appears in any of the terms. Then apply the limit laws. EXAMPLE 2 Find lim f (x) if x→+∞ 282 f (x) = 3x 3 − x . + 7x 2 − 4 2x 3 Curve Sketching and Asymptotes SECTION 4.7 283 Solution We begin by dividing each term in the numerator and denominator by x 3 : 3x 3 − x lim = lim x→+∞ 2x 3 + 7x 2 − 4 x→+∞ 3− 2+ lim x→+∞ = 4 7 − 3 x x 1 x2 3− 2+ lim x→+∞ 1 x2 4 7 − 3 x x = 3 3−0 = . 2+0−0 2 The same computation, but with x → −∞, also gives the result lim f (x) = x→−∞ 3 . 2 ◗ √ √ EXAMPLE 3 Find lim ( x + a − x ). x→+∞ Solution We √ √ use the familiar “divide-and-multiply” technique with the conjugate of x +a − x: √ √ √ √ √ √ x +a + x x + a − x = lim x +a − x · √ lim √ x→+∞ x→+∞ x +a + x = lim √ x→+∞ a √ = 0. x +a + x ◗ Horizontal Asymptotes In geometric terms, the statement lim x→+∞ f (x) = L means that the point (x, f (x)) on the curve y = f (x) approaches the horizontal line y = L as x → +∞. In particular, with the numbers M and of the condition in Eq. (5), the part of the curve for which x > M lies between the horizontal lines y = L − and y = L + (Fig. 4.7.6). Therefore, we say that the line y = L is a horizontal asymptote of the curve y = f (x) if either lim f (x) = L x→+∞ or lim f (x) = L . x→−∞ y y = f (x) y=L+∋ y=L y=L−∋ M x FIGURE 4.7.6 Geometry of the definition of horizontal asymptote. 283 284 CHAPTER 4 Additional Applications of the Derivative EXAMPLE 4 Figure 4.7.7 shows the graph of the function f (x) = 4e2x /(1 + e x )2 . Upon dividing numerator and denominator by e2x , we find that 6 5 y=4 4 f (x) = y 3 y= 2 as x → +∞. Thus the curve y = 4e2x /(1 + e x )2 has the line y = 4 as a horizontal asymptote. ◗ 4e2x (1 + ex)2 1 0 0 1 2 3 x 4e2x 4 = −x → 4 x 2 (1 + e ) (e + 1)2 4 5 6 FIGURE 4.7.7 The graph 4e2x . y= (1 + e x )2 EXAMPLE 5 Figure 4.7.8 shows the graph of the function f (x) = e−x/5 sin 2x. Because |sin 2x| 1 for all x and e−x/5 = 1/(e x/5 ) → 0 as x → +∞, the squeeze law of limits implies that e−x/5 sin 2x → 0 as x → +∞. Thus the curve y = e−x/5 sin 2x ◗ has the x-axis y = 0 as a horizontal asymptote. EXAMPLE 6 Sketch the graph of f (x) = x/(x − 2). Indicate any horizontal or vertical asymptotes. 1 y = e−x/5 sin 2x Solution First we note that x = 2 is a vertical asymptote because | f (x)| → +∞ as x → 2. Also, y = e−x/5 y 0.5 0 lim − 0.5 x→±∞ y = − e−x/5 x = lim x − 2 x→±∞ 1 1− −1 0 5 10 x FIGURE 4.7.8 The graph y = e−x/5 sin 2x. 15 20 2 x = 1 = 1. 1−0 So the line y = 1 is a horizontal asymptote. The first two derivatives of f are f (x) = − 2 (x − 2)2 and f (x) = 4 . (x − 2)3 Neither f (x) nor f (x) is zero anywhere, so the function f has no critical points and no inflection points. Because f (x) < 0 for x = 2, we see that f (x) is decreasing on the open intervals (−∞, 2) and (2, +∞). And because f (x) < 0 for x < 2 and f (x) > 0 for x > 2, the graph of f is concave downward on (−∞, 2) and concave ◗ upward on (2, +∞). The graph of f appears in Fig. 4.7.9. y y= Horizontal asymptote y=1 x x−2 x Vertical asymptote x=2 FIGURE 4.7.9 The graph for Example 6. EXAMPLE 7 Let’s reexamine the function f (x) = 284 x (x + 2)2 Curve Sketching and Asymptotes SECTION 4.7 285 whose graph was shown in Fig. 4.7.2. We note that (2, 18 ) (4, 19 ) 0.2 0.1 y y= x lim = lim x→∞ (x + 2)2 x→∞ x (x + 2)2 0 −0.1 −0.2 −10 0 10 20 x 30 40 50 x → 0 as (x + 2)2 x → ∞, so the x-axis y = 0 is a horizontal asymptote. 1 x 2 1+ x 2 = 0, so the x-axis y = 0 is a horizontal asymptote of the graph y = f (x). We must change the viewing window to see clearly the behavior of this curve for x > 0. With the window −10 < x < 40, −0.25 < y < 0.25 of Fig. 4.7.10 we see that f (x) appears to attain a local maximum value near the point where x = 2 before approaching zero as x → ∞. Indeed, upon differentiating f and simplifying the result, we see that f (x) = FIGURE 4.7.10 2−x , (x + 2)3 so the indicated maximum point on the curve is (2, 18 ). The second derivative of f is f (x) = 2(x − 4) , (x + 2)4 and it follows that the inflection point apparent in Fig. 4.7.10 is at (4, 19 ). ◗ Curve-Sketching Strategy The curve-sketching techniques of Sections 4.5 and 4.6, together with those of this section, can be summarized as a list of steps. If you follow these steps, loosely rather than rigidly, you will obtain a qualitatively accurate sketch of the graph of a given function f : 1. Solve the equation f (x) = 0 and also find where f (x) does not exist. This gives the critical points of f . Note whether the tangent line is horizontal, vertical, or nonexistent at each critical point. 2. Determine the intervals on which f is increasing and those on which it is decreasing. 3. Solve the equation f (x) = 0 and also find where f (x) does not exist. These points are the possible inflection points of the graph. 4. Determine the intervals on which the graph of f is concave upward and those on which it is concave downward. 5. Find the y-intercept and the x-intercepts (if any) of the graph. 6. Plot and label the critical points, possible inflection points, and intercepts. 7. Determine the asymptotes (if any), discontinuities (if any), and especially the behavior of f (x) and f (x) near discontinuities of f . Also determine the behavior of f (x) as x → +∞ and as x → −∞. 8. Finally, join the plotted points with a curve that is consistent with the information you have gathered. Remember that corner points are rare and that straight sections of graph are even rarer. You may follow these steps in any convenient order and omit any that present major computational difficulties. Many problems require fewer than all eight steps; see Example 6. But Example 8 requires them all. EXAMPLE 8 Sketch the graph of f (x) = 2 + x − x2 . (x − 1)2 Solution We notice immediately that lim f (x) = +∞, x→1 285 286 CHAPTER 4 Additional Applications of the Derivative because the numerator approaches 2 as x → 1 and the denominator approaches zero through positive values. So the line x = 1 is a vertical asymptote. Also, 2 1 + −1 2 2 + x − x2 x x = lim = −1, lim x→±∞ (x − 1)2 x→±∞ 1 2 1− x so the line x = −1 is a horizontal asymptote (in both the positive and the negative directions). Next we apply the quotient rule and simplify to find that f (x) = x −5 . (x − 1)3 Thus the only critical point in the domain of f is x = 5, and we plot the point (5, f (5)) = (5, − 98 ) on a convenient coordinate plane and mark the horizontal tangent there. To determine the increasing or decreasing behavior of f , we use both the critical point x = 5 and the point x = 1 (where f is not defined) to separate the x-axis into open intervals. The following table shows the results: Interval (x − 1)3 x −5 f (x) (−∞, 1) (1, 5) (5, +∞) − + + − − + + − + f Increasing Decreasing Increasing After some simplifications, we find the second derivative to be f (x) = 2(7 − x) . (x − 1)4 ) on The only possible inflection point is at x = 7, corresponding to the point (7, − 10 9 the graph. We use both x = 7 and x = 1 (where f is undefined) to separate the x-axis into open intervals. The concave structure of the graph can be deduced with the aid of the next table. Interval (x − 1)4 7−x f (x) (−∞, 1) (1, 7) (7, +∞) + + + + + − + + − f Concave upward Concave upward Concave downward The y-intercept of f is (0, 2), and the equation 2 + x − x 2 = 0 readily yields the x-intercepts (−1, 0) and (2, 0). We plot these intercepts, sketch the asymptotes, and finally sketch the graph with the aid of the two tables; their information now is ◗ symbolized along the x-axis in Fig. 4.7.11. Slant Asymptotes Not all asymptotes are horizontal or vertical—some are inclined. The nonvertical line y = mx + b is an asymptote for the curve y = f (x) provided that either lim [ f (x) − (mx + b)] = 0 (7a) lim [ f (x) − (mx + b)] = 0 (7b) x→+∞ or x→−∞ (or both). These conditions mean that as x → +∞ or as x → −∞ (or both), the vertical distance between the point (x, f (x)) on the curve and the point (x, mx + b) on the line approaches zero. 286 Curve Sketching and Asymptotes SECTION 4.7 287 y 2 y = 2 + x − x2 (x − 1) (0, 2) y -intercept (−1, 0) x-intercept + + + (2, 0) x -intercept + − − − − + + (5, − 98 ) x = 1: vertical asymptote local and global minimum + + + + + + + + x y = −1: horizontal asymptote ) (7, − 10 9 inflection point FIGURE 4.7.11 Graphing the function of Example 8. Suppose that f (x) = p(x)/q(x) is a rational function for which the degree of p(x) is greater by 1 than the degree of q(x). Then, by long division of q(x) into p(x), we find that f (x) has the form f (x) = mx + b + g(x) where m = 0 and lim g(x) = 0. x→±∞ Thus the nonvertical line y = mx + b is an asymptote of the graph of y = f (x). Such an asymptote is called a slant asymptote. EXAMPLE 9 Sketch the graph of f (x) = x2 + x − 1 . x −1 Solution The long division suggested previously takes the form shown next. x +2 x − 1 ) x2 + x − 1 x2 − x 2x − 1 2x − 2 1 Thus f (x) = x + 2 + 1 . x −1 So y = x + 2 is a slant asymptote of the curve. Also, lim | f (x)| = +∞, x→1 so x = 1 is a vertical asymptote. The first two derivatives of f are f (x) = 1 − 1 x(x − 2) = 2 (x − 1) (x − 1)2 287 288 CHAPTER 4 Additional Applications of the Derivative and f (x) = 2 . (x − 1)3 It follows that f has critical points at x = 0 and at x = 2 but no inflection points. The sign of f tells us that f is increasing on (−∞, 0) and on (2, +∞), decreasing on (0, 1) and on (1, 2). Examination of f (x) reveals that f is concave downward on (−∞, 1) and concave upward on (1, +∞). In particular, f (0) = 1 is a local maximum value and f (2) = 5 is a local minimum value. The graph of f looks much like the one ◗ in Fig. 4.7.12. y y = x + 2: asymptote (0, 1) Local maximum y-intercept (2, 5) Local minimum x = 1: vertical asymptote x 2 y= x +x−1 x−1 20 FIGURE 4.7.12 A function with slant asymptote y = x + 2 (Example 9). 15 10 y 5 Calculator/Computer Graphing 0 Instead of using concepts of calculus to construct a graph from scratch, we can go the other way. That is, we can begin with a graph plotted by a calculator or computer, and then use a calculator to analyze the graph and refine our understanding of it. In Sections 1.3 and 1.4 we discussed the fact that a calculator or computer graph can sometimes be misleading or incomplete. But now we can use calculus—and in particular the computation of critical points and inflection points—to make sure that the machinegenerated graph exhibits all of its important features. Moreover, with graphing and automatic solution techniques we can investigate graphs of functions that would be too complicated to study without a calculator or computer. −5 −10 −4 0 x 4 FIGURE 4.7.13 y = x 4 − 5x 2 − 5x + 7 . 2x 3 − 2x + 1 EXAMPLE 10 Figure 4.7.13 shows a computer-generated graph of the function x 4 − 5x 2 − 5x + 7 . (8) 2x 3 − 2x + 1 It appears likely to have a vertical asymptote somewhere near x = −1. To test this hypothesis, we need to know where the denominator in (8) is zero. The graph of this denominator, shown in Fig. 4.7.14, indicates that the equation 2x 3 − 2x + 1 = 0 has a single real solution near x = −1.2. We could zoom in graphically to show that the corresponding vertical asymptote is still closer to x = −1.19, and a calculator or computer Solve command yields the solution x ≈ −1.1915 accurate to four decimal places. Noting that the degree of the numerator in (8) exceeds that of the denominator, we find by long division that f (x) = 5 4 3 3 2 y = 2x − 2x + 1 1 y 0 −1 −2 −3 −4 −5 −2 −1 0 x 1 FIGURE 4.7.14 Graph of the denominator in (8). 288 2 f (x) = x +7 −4x 2 − 11 1 2 x+ . 3 2 2x − 2x + 1 Thus the graph y = f (x) has the slant asymptote y = 12 x (Fig. 4.7.15). Curve Sketching and Asymptotes SECTION 4.7 To investigate the critical points of f (x), we calculate the derivative 15 f (x) = 10 y y= 1 x 2 0 −4 −2 2 4 x has four real solutions, near the points x = −2.3, −0.6, 0.5, and 1.1. We could zoom in closer to each of these solutions, or use a calculator or computer Solve command to get the approximations x ≈ −2.3440, −0.5775, 0.4673, and 1.0864 that agree with the overall structure of the graph shown in Fig. 4.7.13, where four critical points with horizontal tangent lines are apparent. The leftmost critical point x ≈ −2.3440 deserves closer examination. In Fig. 4.7.15 it appears likely to lie just to the left of the point where the left branch of the graph y = f (x) crosses the slant asymptote y = 12 x. The zoom shown in Fig. 4.7.17 bears out this observation. Finally, an examination of the original graph y = f (x) in Fig. 4.7.13 suggests the approximate locations of three inflection points in the first quadrant. But if the graph is to approach the slant asymptote as x → −∞, then Fig. 4.7.17 suggests the presence of a fourth inflection point somewhere to the left of the leftmost critical point. (Why?) To investigate this possibility, we calculate the second derivative 60 40 20 0 −20 −40 −60 −80 −100 −120 − 4 −3 −2 −1 (9) 2x 6 + 4x 4 + 24x 3 − 32x 2 − 10x + 9 = 0 0 FIGURE 4.7.15 Now we see both the vertical asymptote and the slant asymptote y = 12 x. y 2x 6 + 4x 4 + 24x 3 − 32x 2 − 10x + 9 . (2x 3 − 2x + 1)2 The critical points of f (x) are the zeros of the numerator of f (x), together with the zero of the denominator that yields the vertical asymptote. The graph of the numerator, shown in Fig. 4.7.16, indicates that the equation 5 −5 −6 289 0 x 1 2 3 4 f (x) = 2(−16x 6 − 66x 5 + 120x 4 + 34x 3 − 18x 2 − 42x + 13) . (2x 3 − 2x + 1)3 (10) The inflection points of y = f (x) have x-coordinates given by the zeros of the numerator in (10). The graph of this numerator, shown in Fig. 4.7.18, indicates that the equation FIGURE 4.7.16 Graph of the numerator in (9). 2(−16x 6 − 66x 5 + 120x 4 + 34x 3 − 18x 2 − 42x + 13) = 0 has four real solutions—a negative one near −5.5 as well as three positive solutions between 0 and 2 that correspond to the visually apparent first-quadrant inflection points in Fig. 4.7.13. We could zoom in closer to each of these solutions, or use a calculator or computer Solve command to get the approximations −5.4303, 0.3152, 0.6503, and 1.3937. The larger view shown in Fig. 4.7.19 convinces us that we’ve found all the inflection points of y = f (x). In particular, we see that y = f (x) is concave upward to the left of the inflection point x ≈ −5.4303, where the denominator in (10) is negative (why?), and is concave downward just to its right (consistent with what we see in Fig. 4.7.17). − 0.5 y= 1 x 2 −1 y −1.5 −2 −3.5 y = f(x) −3 −2.5 x −2 −1.5 FIGURE 4.7.17 Near the leftmost critical point. 100 80 60 40 20 y 0 −20 − 40 −60 −80 −100 −8 10 × 10 4 8 6 y 4 2 0 −2 −6 −4 −2 x 0 FIGURE 4.7.18 Graph of the numerator in (10). 2 4 −4 −8 −6 −4 −2 x 0 2 4 FIGURE 4.7.19 Larger view of the graph of the numerator in (10). 289 290 CHAPTER 4 Additional Applications of the Derivative This thorough analysis of the graph of the function f of Eq. (8) involves a certain amount of manual labor—just to calculate and simplify the derivatives in (9) and (10) unless we use a computer algebra system for this task—but would be very challenging ◗ without the use of a graphing calculator or computer. 4.7 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If the function f increases without bound as x → a, then lim f (x) = +∞. x→a x 2. lim = +∞. x→−2 (x + 2)2 1 has a vertical asymptote with equation x = 1. 3. The graph of f (x) = x −1 2 3x 3 − x = . 4. lim 3 2 x→∞ 2x + 7x − 4 3 5. The line y = L is a horizontal asymptote of f (x) if lim f (x) = +∞. x→L 6. 7. 8. 9. 10. x has no horizontal asymptote. The graph of f (x) = (x + 2)2 2 + x − x2 x −5 If f (x) = , then f (x) = . (x − 1)2 (x − 1)3 The graph of x2 + x − 1 f (x) = x −1 has only one asymptote and exactly two extrema. x2 + x − 1 2 If f (x) = . , then f (x) = x −1 (x − 1)3 A graph cannot have both a vertical asymptote and a slant asymptote. 4.7 CONCEPTS: QUESTIONS AND DISCUSSION 1. Can you sketch the graph of a function that has two distinct critical points that are not separated by an inflection point? Does such a function exist? 2. Does the graph of a polynomial always have an inflection point? Or does it depend upon whether the degree n of the polynomial is even or odd? Begin by discussing separately the cases n = 2, 3, 4, and 5. 3. Can the graph of a polynomial have an asymptote? Does the graph of a rational function always have an asymptote? Justify your answers. 4. What can you say about the degrees of the numerator and denominator of a rational function that has a horizontal asymptote? What can you say about the degrees of the numerator and denominator of a rational function that has a slant asymptote? 4.7 PROBLEMS Investigate the limits in Problems 1 through 16. x 1. lim x→+∞ x + 1 x2 + x − 2 3. lim x→1 x −1 2x 2 − 1 5. lim x→+∞ x 2 − 3x 290 x2 + 1 2. lim x→−∞ x 2 − 1 x2 − x − 2 4. lim x→1 x −1 1 + ex 6. lim x→+∞ 2 + e 2x 7. lim x→−1 x 2 + 2x + 1 (x + 1)2 8. x −4 9. lim √ x→4 x −2 10. √ 8− 3 x 11. lim x→−∞ 2 + x 12. lim 5x 3 − 2x + 1 7x 3 + 4x 2 − 2 lim 2x + 1 √ x−x x lim 4e6x + 5 sin 6x (1 + 2e2x )3 x→+∞ x→+∞ x→+∞ Curve Sketching and Asymptotes SECTION 4.7 √ 3 4x 2 − x 13. lim x→+∞ x2 + 9 x 2 + 2x − x 15. lim x 3 − 8x + 1 14. lim x→−∞ 3x − 4 16. lim 2x − 4x 2 − 5x x→−∞ x→−∞ Apply your knowledge of limits and asymptotes to match each function in Problems 17 through 28 with its graph-withasymptotes in one of the twelve parts of Fig. 4.7.20. 1 1−x 17. f (x) = 1 x −1 18. f (x) = 19. f (x) = 1 (x − 1)2 20. f (x) = − 1 (1 − x)2 1 1 − x2 x 24. f (x) = 1 − x2 1 x2 − 1 x 23. f (x) = 2 x −1 22. f (x) = 21. f (x) = −4 0 x 26. f (x) = x2 x2 − 1 27. f (x) = x2 x −1 28. f (x) = x3 x2 − 1 Sketch by hand the graph of each function in Problems 29 through 54. Identify and label all extrema, inflection points, intercepts, and asymptotes. Show the concave structure clearly as well as the behavior of the graph for |x| large and for x near any discontinuities of the function. 29. f (x) = 2 x −3 30. f (x) = 31. f (x) = 3 (x + 2)2 32. f (x) = − 33. f (x) = 1 (2x − 3)3 34. f (x) = 4 2 2 y 0 y 0 −2 −2 −4 −4 −4 (a) x x −1 4 4 y 0 25. f (x) = −4 4 −2 (b) 0 x 2 4 −4 −2 0 x 2 4 −4 −2 0 x 2 4 −4 −2 0 x 2 4 (c) 4 5−x 4 (3 − x)2 x +1 x −1 4 4 4 291 2 y 0 y 0 −4 −4 y 0 −2 −4 0 x (d) −4 −4 4 0 x (e) 4 (f) 4 4 4 2 2 2 y 0 y 0 y 0 −2 −2 −2 −4 −4 −4 −4 −2 (g) 0 x 2 −4 4 (h) 4 −2 0 x 2 4 (i) 4 4 y 0 y 0 −4 −4 2 y 0 −2 −4 −4 (j) −2 0 x 2 −4 4 (k) 0 x −4 4 (l) 0 x 4 FIGURE 4.7.20 Problems 17 through 28. 291 292 CHAPTER 4 Additional Applications of the Derivative x2 x2 + 1 1 37. f (x) = 2 x −9 35. f (x) = 1 x2 + x − 6 1 41. f (x) = x + x 39. f (x) = x2 x −1 1 45. f (x) = (x − 1)2 43. f (x) = ex +1 1 49. f (x) = 2 x −x −2 47. f (x) = ex 2x 2 x +1 x 38. f (x) = 4 − x2 (x + 1)2 (x − 3)2 (x + 1)(x − 3)4 60. f (x) = x 3 (x − 2)2 x 3 (x − 2)3 36. f (x) = 59. f (x) = 40. f (x) = In Problems 61 through 68, begin with a calculator- or computergenerated graph of the curve y = f (x). Then use a calculator or computer to locate accurately the vertical asymptotes and the critical and inflection points of f (x). Finally, use a calculator or computer to produce graphs that display the major features of the curve, including any vertical, horizontal, and slant asymptotes. 2x 2 + 1 x 2 − 2x 42. f (x) = 2x + e−x 2x 3 − 5x 2 + 4x x 2 − 2x + 1 1 46. f (x) = (1 + e x )2 44. f (x) = 1 + e−x 1 50. f (x) = (x − 1)(x + 1)2 48. f (x) = ex 51. f (x) = x2 − 4 x 52. f (x) = e x − e−x e x + e−x 53. f (x) = x3 − 4 x2 54. f (x) = x2 + 1 x −2 In Problems 55 through 60, you can determine by inspection the x-intercepts as well as the vertical and horizontal asymptotes of the curve y = f (x). First sketch the graph by hand, using this information, and without calculating any derivatives. Then use a calculator or computer to locate accurately the critical and inflection points of f (x). Finally, use a calculator or computer to produce graphs that display the major features of the curve. 55. f (x) = (x + 1)(x − 3) x 2 (x − 2) 56. f (x) = (x + 1)2 (x − 3) x 2 (x − 4) 57. f (x) = (x + 1)2 (x − 3) x 3 (x − 2) 58. f (x) = (x + 1)2 (x − 3)2 x 3 (x − 2) 61. f (x) = x2 x 3 − 3x 2 + 1 62. f (x) = x2 x 3 − 3x 2 + 5 63. f (x) = x 4 − 4x + 5 x 3 − 3x 2 + 5 64. f (x) = x 4 − 4x + 1 2x 3 − 3x + 2 65. f (x) = x 5 − 4x 2 + 1 2x 4 − 3x + 2 66. f (x) = x 5 − 4x 3 + 2 2x 4 − 5x + 5 67. f (x) = x 6 − 4x 3 + 5x 2x 5 − 5x 3 + 5 68. f (x) = 2x 6 − 5x 4 + 6 3x 5 − 5x 4 + 4 69. Suppose that f (x) = x 2 + 2 . x Note that lim [ f (x) − x 2 ] = 0, x→±∞ so the curve y = f (x) approaches the parabola y = x 2 as x → ±∞. Use this observation to make an accurate sketch of the graph of f . 70. Use the method of Problem 69 to make an accurate sketch of the graph of f (x) = x 3 − 12 . x −1 4.7 INVESTIGATION: Locating Special Points on Exotic Graphs The investigations described here deal with fairly exotic curves having critical and inflection points that are not clearly visible on their graphs if plotted on a “natural” scale. The reason is that different scales on the x- and y-axes are required to see the unusual behavior in question. In both investigations you are to begin with a graph that you generate with calculator or computer, and then analyze the curve—locating accurately all critical and inflection points—in order to plot additional graphs that demonstrate clearly all of the major features of the curve. Investigation A Choose in advance a single-digit integer n (perhaps the final nonzero digit of your student I.D. number). Then your task is to analyze the structure of the curve y = x 7 + 5x 6 − 11x 5 − 21x 4 + 31x 3 − 57x 2 − (101 + 2n)x + (89 − 3n). Find the local maximum and minimum points and the inflection point (or points) on this curve, giving their coordinates accurate to four decimal places. To display all these points, you probably will need to produce separate plots with different scales, showing different parts of this curve. In the end, use all the information accumulated to produce a careful hand sketch (not to scale) displaying all the maxima, minima, and inflection points with their (approximate) coordinates labeled. 292 ^ Indeterminate Forms and L'Hopital's Rule SECTION 4.8 293 Investigation B Explore in the detail the structure of the graph of the function f (x) = − 1,234,567,890 + 2,695,140,459x 2 + 605,435,400x 3 − 411,401,250x 4 − 60,600,000x 5 + 25,000,000x 6 . The graph y = f (x) is shown in Fig. 4.7.21. At a glance, it might appear that we have only three critical points—a local minimum near the origin and two critical points that are also inflection points, as well as two more inflection points that are not critical points. Settle the matter. How many of each, in fact, are there? Find and exhibit all of them in a graph; your graph may be a neat hand sketch and need not be to scale. 25 × 10 9 20 15 y 10 5 0 −5 −5 0 x 5 FIGURE 4.7.21 The ”big picture” in Investigation B. ^ 4.8 INDETERMINATE FORMS AND L'HOPITAL'S RULE An indeterminate form is a certain type of expression with a limit that is not evident by inspection. There are several types of indeterminate forms. If lim f (x) = 0 = lim g(x), x→a x→a then we say that the quotient f (x)/g(x) has the indeterminate form 0/0 at x = a (or as x → a). For example, to differentiate the trigonometric functions (Section 3.7), we needed to know that lim x→0 2 1.5 1 y 0.5 (0, 1) y = sin x x 0 −0.5 −1 −8 −6 −4 −2 0 2 4 6 8 x FIGURE 4.8.1 Visual evidence that the quotient (sin x)/x is near 1 when x is near zero. sin x = 1. x (1) Figure 4.8.1 corroborates the fact that (sin x)/x is close to 1 when x is close to zero. The quotient (sin x)/x in Eq. (1) has the indeterminate form 0/0 at x = 0 because the functions f (x) = sin x and g(x) = x both approach zero as x → 0. Hence the quotient law of limits cannot be used to evaluate this limit. We therefore needed a special geometric argument (see Section 2.3) to find the limit in Eq. (1). Something similar happens whenever we compute a derivative, because the quotient f (x) − f (a) , x −a whose limit as x → a is the derivative f (a), has the indeterminate form 0/0 at x = a. We can sometimes find the limit of an indeterminate form by performing a special algebraic manipulation or construction, as in our earlier computation of derivatives. Often, however, it is more convenient to apply a rule that appeared in the first calculus textbook ever published, by the Marquis de l’Hôpital, in 1696. L’Hôpital was a French nobleman who had hired the Swiss mathematician John Bernoulli as his calculus tutor, and “l’Hôpital’s rule” is actually the work of Bernoulli. 293 294 CHAPTER 4 Additional Applications of the Derivative ^ THEOREM 1 L'Hopital's Rule Suppose that the functions f and g are differentiable and that g (x) is nonzero in some neighborhood of the point a (except possibly at a itself). Suppose also that lim f (x) = 0 = lim g(x). x→a x→a Then lim x→a f (x) f (x) = lim , x→a g (x) g(x) (2) provided that the limit on the right either exists (as a finite real number) or is +∞ or −∞. In essence, l’Hôpital’s rule says that if f (x)/g(x) has the indeterminate form 0/0 at x = a, then—subject to a few mild restrictions—this quotient has the same limit at x = a as does the quotient f (x)/g (x) of derivatives. The proof of l’Hôpital’s rule is discussed at the end of this section. 3 2.5 2 1.5 y 1 0.5 EXAMPLE 1 Find lim (0,, 12 ) x y= e −1 sin 2x 0 −0.5 −1 −1.5 −1 − 0.5 0 x 0.5 1 1.5 FIGURE 4.8.2 Visual evidence that ex − 1 is near 12 when the quotient sin 2x x is near 0. x→0 ex − 1 . sin 2x Solution The fraction whose limit we seek has the indeterminate form 0/0 at x = 0. The numerator and denominator are clearly differentiable in some neighborhood of x = 0, and the derivative of the denominator is certainly nonzero if the neighborhood is small enough (specifically, if |x| < π/4). So l’Hôpital’s rule applies, and lim x→0 ex − 1 ex e0 1 = lim = = x→0 2 cos 2x sin 2x 2 cos 0 2 because (by continuity) both e x and cos 2x approach 1 as x → 0. Figure 4.8.2 corrob◗ orates this limit. If the quotient f (x)/g (x) is itself indeterminate, then l’Hôpital’s rule may be applied a second (or third, . . . ) time, as in Example 2. When the rule is applied repeatedly, however, the conditions for its applicability must be checked at each stage. EXAMPLE 2 Find lim x→1 1 − x + ln x . 1 + cos π x Solution 1 −1 + 1 − x + ln x x = lim lim x→1 1 + cos π x x→1 −π sin π x x −1 = lim x→1 π x sin π x = lim x→1 =− (still of the form 0/0) (algebraic simplification) 1 π sin π x + π 2 x cos π x 1 π2 (l’Hôpital’s rule again) (by inspection). Because the final limit exists, so do the previous ones; the existence of the final limit in Eq. (2) implies the existence of the first. ◗ When you need to apply l’Hôpital’s rule repeatedly in this way, you need only keep differentiating the numerator and denominator separately until at least one of them has a nonzero finite limit. At that point you can recognize the limit of the quotient by inspection, as in the final step in Example 2. EXAMPLE 3 Find lim x→0 294 sin x . x + x2 ^ Indeterminate Forms and L'Hopital's Rule SECTION 4.8 295 Solution If we simply apply l’Hôpital’s rule twice in succession, the result is the incorrect computation lim x→0 sin x cos x = lim x→0 1 + 2x x + x2 − sin x = lim = 0. x→0 2 (Wrong!) The answer is wrong because (cos x)/(1 + 2x) is not an indeterminate form. Thus l’Hôpital’s rule cannot be applied to it. The correct computation is lim x→0 lim cos x 1 sin x cos x x→0 = = = 1. = lim x→0 1 + 2x x + x2 lim (1 + 2x) 1 ◗ x→0 The point of Example 3 is to issue a warning: Verify the hypotheses of l’Hôpital’s rule before you apply it. It is an oversimplification to say that l’Hôpital’s rule works when you need it and doesn’t work when you don’t, but there is still much truth in this statement. Indeterminate Forms Involving ∞ L’Hôpital’s rule has several variations. In addition to the fact that the limit in Eq. (2) is allowed to be infinite, the real number a in l’Hôpital’s rule may be replaced with either +∞ or −∞. For example, lim x→∞ f (x) f (x) = lim x→∞ g (x) g(x) (3) provided that the other hypotheses are satisfied in some interval of the form (c, +∞). In particular, to use Eq. (3), we must first verify that lim f (x) = 0 = lim g(x) x→∞ x→∞ and that the right-hand limit in Eq. (3) exists. The proof of this version of l’Hôpital’s rule is outlined in Problem 70. L’Hôpital’s rule may also be used when f (x)/g(x) has the indeterminate form ∞/∞. This means that y 8 lim f (x) is either +∞ or −∞ x→a 6 and 4 y= 2 1 2 ln 2x ln x 3 lim g(x) x→a 4 x −2 −4 FIGURE 4.8.3 The graph ln 2x has both the vertical y= ln x asymptote x = 1 and the horizontal asymptote y = 1. is either +∞ or −∞. The proof of this extension of the rule is difficult and is omitted here. [For a proof, see (for example) A. E. Taylor and W. R. Mann, Advanced Calculus, 3rd ed. (New York: John Wiley, 1983), p. 107. ] One-sided indeterminate forms occur, and we may speak of a 0/0 form or an ∞/∞ form as either x → a − or as x → a + . EXAMPLE 4 Figure 4.8.3 shows a computer-generated graph of the function ln 2x . (4) ln x The vertical asymptote x = 1 is explained (without using l’Hôpital’s rule) by the facts that f (x) = • • the numerator ln 2x is positive at x = 1, while the denominator ln x approaches zero through negative values as x → 1− and approaches zero through positive values as x → 1+ . 295 296 CHAPTER 4 Additional Applications of the Derivative y = ex y (v, u) y = ln x 1 The familiar graph of y = ln x (Fig. 4.8.4) reminds us that ln x → −∞ as x → 0+ and that ln x → +∞ as x → +∞. Consequently, we see that the function f in Eq. (4) has the indeterminate form ∞/∞, both as x → 0+ and as x → +∞. Thus l’Hôpital’s rule gives 2 ln 2x = lim 2x = lim 1 = 1 lim x→0+ ln x x→0+ 1 x→0+ x (u, v) 1 x y=x and 2 ln 2x 2x = lim = lim 1 = 1. lim x→+∞ ln x x→+∞ 1 x→+∞ x FIGURE 4.8.4 The graphs y = e x and y = ln x are reflections of one another in the line y = x. The fact that ln 2x =1 ln x lim x→0+ explains why the graph in Fig. 4.8.3 appears to “start” at the point (0, 1). And the fact that lim x→∞ ln 2x =1 ln x explains the horizontal asymptote y = 1 that we see in Fig. 4.8.3. ◗ Order of Magnitude of e x and ln x Because Dx e x = e x (for all x) and Dx ln x = 1/x (for all x > 0), the functions ln x and e x are increasing wherever they are defined. If n is an integer and x > n, it follows that e x > en > 2n , and hence that lim e x = +∞. x→+∞ (5) Similarly, if x > 2n , then ln x > ln 2n = n ln 2, and therefore lim ln x = +∞ x→+∞ (6) as well. But the graphs in Fig. 4.8.4 suggest that as x → +∞, e x → +∞ much more rapidly than ln x → +∞. Indeed, l’Hôpital’s rule yields lim x→+∞ ex ex = lim = lim xe x = +∞. x→+∞ 1 x→+∞ ln x x Thus, when x is large positive, e x is so much larger than ln x that the quotient e x /(ln x) is large positive. This observation seems related to the facts that: • • 296 The second derivative Dx2 e x = e x > 0 for all x, so the curve y = e x is concave upward, and becomes steeper and steeper as x increases. In contrast, Dx2 ln x = −1/x 2 < 0 for all x > 0, so the curve y = ln x is concave downward, and becomes less and less steep as x increases. ^ Indeterminate Forms and L'Hopital's Rule SECTION 4.8 297 EXAMPLE 5 Explain the principal features of the graph of the function f (x) = x 2 e−x shown in Fig. 4.8.5. y 0.6 Solution The function f (x) = x 2 /e x has the indeterminate form ∞/∞ as x → +∞. Hence the horizontal asymptote y = 0 is explained by two applications of l’Hôpital’s rule: 0.4 y= 0.2 2 4 6 x2 ex 8 FIGURE 4.8.5 The graph y = x 2 e−x has two local extrema, two inflection points, and the horizontal asymptote y = 0. lim x x→∞ x2 2x 2 = lim x = lim x = 0. x→∞ e x→∞ e ex (7) The two local extrema that we see in the figure result from the fact that the derivative f (x) = 2x · e−x − x 2 · e−x = (2x − x 2 )e−x = x(2 − x)e−x has the two zeros x = 0 and x = 2. The second derivative is f (x) = (2 − 2x) · e−x − (2x − x 2 ) · e−x = (x 2 − 4x + 2)e−x , √ and the two solutions x = 2 ± 2 of the quadratic equation x 2 − 4x + 2 = 0 provide ◗ the two inflection points visible in Fig. 4.8.5. The exponential function is notable for its very rapid rate of increase with increasing x. In fact, e x increases more rapidly as x → +∞ than any fixed power of x. Thus the limit in (7) is a special case of the fact that lim xk =0 ex (8) ex = +∞ xk (9) x→∞ or, alternatively, lim x→∞ for any fixed real number k > 0. For instance, if k = n, a positive integer, then n successive applications of l’Hôpital’s rule give lim x→∞ xn nx n−1 n(n − 1)x n−2 = lim = lim x→∞ x→∞ ex ex ex = · · · = lim x→∞ n(n − 1) · · · 2x n(n − 1) · · · 2 · 1 = lim = 0. x x→∞ e ex In Problem 61 we ask you to consider positive nonintegral values of k. The table in Fig. 4.8.6 illustrates the case k = 5 of Eq. (8). Although both x 5 → +∞ and e x → +∞ as x → +∞, we see that e x increases so much more rapidly than x 5 that x 5 /e x → 0. x x5 ex x 5 /e x 10 20 30 40 50 1.00 × 105 3.20 × 106 2.43 × 107 1.02 × 108 3.13 × 108 ↓ ∞ 2.20 × 104 4.85 × 108 1.07 × 1013 2.35 × 1017 5.18 × 1021 ↓ ∞ 4.54 × 100 6.60 × 10−3 2.27 × 10−6 4.35 × 10−10 6.03 × 10−14 ↓ ∞ FIGURE 4.8.6 Orders of magnitude of x 5 and e x . 297 298 CHAPTER 4 Additional Applications of the Derivative EXAMPLE 6 Explain the principal features of the graph of the function y 1 ln x f (x) = √ x y = ln x √x 0.5 shown in Fig. 4.8.7. 20 40 60 80 x − 0.5 √ Solution The function f (x) = (ln x)/ x has the indeterminate form ∞/∞ as x → +∞. A single application of l’Hôpital’s rule yields 1 x 1 √ −1 ln x lim √ = lim x→∞ x→∞ x FIGURE 4.8.7 √ The graph y = (ln x)/ x has a local maximum, an inflection point, and the horizontal asymptote y = 0. 2 = lim √ = 0, x→∞ x (10) 2 x and thus the graph has the horizontal asymptote y = 0. The local maximum that we see in the figure results from the fact that the derivative 1√ ln x x − √ x 2 x f (x) = x = 2 − ln x 2x 3/2 has the zero x = e2 . The inflection point that we see corresponds to the zero x = e8/3 of the second derivative 1 − · 2x 3/2 − (2 − ln x) 3x 1/2 −8 + 3 ln x f (x) = x = . ◗ 3 4x 4x 5/2 In contrast with the exponential function, the natural logarithm function is notable for its very slow rate of increase with increasing x. In Problem 62 we ask you to generalize the result in Eq. (10) by showing that 4 3 2 y = ln x lim x→+∞ y 1 y = x1/10 0 −1 • FIGURE 4.8.8 Comparing y = ln x with y = x 1/10 . (11) if k > 0. Thus • 1 2 3 4 5 6 7 8 9 x ln x =0 xk ln x increases slower than any (positive) power of x, whereas e x increases faster than any power of x. REMARK Figure 4.8.8 might suggest to the unwary that ln x is greater than (rather than less than) x 1/10 when x is large positive. But Eq. (11) implies that the graph of y = x 1/10 must eventually overtake and recross the graph of y = ln x. (See Problem 74.) ^ Proof of L'Hopital's Rule Suppose that the functions f and g of Theorem 1 are not merely differentiable but have continuous derivatives near x = a and that g (a) = 0. Then lim f (x) f (a) f (x) x→a = = lim x→a g (x) lim g (x) g (a) (12) x→a by the quotient law for limits. In this case l’Hôpital’s rule in Eq. (2) reduces to the limit lim x→a f (a) f (x) = , g(x) g (a) (13) which is a weak form of the rule. It actually is this weak form that is typically applied in single-step applications of l’Hôpital’s rule. 298 ^ Indeterminate Forms and L'Hopital's Rule SECTION 4.8 299 EXAMPLE 7 In Example 1 we had f (x) = e x − 1, g(x) = sin 2x so f (x) = e x , g (x) = 2 cos 2x, and g (0) = 2 = 0. With a = 0, Eq. (13) therefore gives lim x→0 ex − 1 f (x) f (0) 1 = lim = = . x→0 g(x) sin 2x g (0) 2 ◗ ^ THEOREM 2 L'Hopital's Rule (weak form) Suppose that the functions f and g are differentiable at x = a, that f (a) = 0 = g(a), and that g (a) = 0. Then lim x→a f (a) f (x) = . g(x) g (a) (13) Proof We begin with the right-hand side of Eq. (13) and work toward the left-hand side. f (a) = g (a) y Slope: f '(a) g' (a) P(g(t), f(t)) Slope: O f (x) − f (a) x −a g(x) − g(a) lim x→a x −a lim x→a = lim f (t) g(t) x→a x = lim f (x) − f (a) g(x) − g(a) = lim f (x) g(x) x→a FIGURE 4.8.9 Suppose that the point P(g(t), f (t)) traces a continuous curve that passes through the origin O when t = a. Then the secant line O P approaches the tangent line at O as t → a, so its slope f (t)/g(t) approaches the slope f (a)/g (a) of the tangent line at O. f (x) − f (a) x −a g(x) − g(a) x −a x→a (the definition of the derivative) (the quotient law of limits) (algebraic simplification) [because f (a) = 0 = g(a)]. ◆ Figure 4.8.9 illustrates the meaning and proof of Theorem 2. Appendix H includes a proof of the strong form of l’Hôpital’s rule, the form stated in Theorem 1. 4.8 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If lim f (x) = 0 = lim g(x), then x→a 0 at x = a. 0 sin x 2. lim = 1. x→0 x 1 ex − 1 = . 4. lim x→0 sin 2x 2 ln 2x = 1. 6. lim x→∞ ln x x→a f (x) is said to have the indeterminate form g(x) ex − 1 0 = = 1. x→0 sin 2x 0 sin x 5. lim = 0. x→0 x + x 2 ln x ∞ 7. lim √ = = 1. x→∞ ∞ x 3. lim 299 300 CHAPTER 4 Additional Applications of the Derivative ln x 2 1 − x + ln x 9. lim does not exist. 8. lim √ = lim √ = 0. x→∞ x→∞ x→1 1 + cos π x x x 10. If f and g are differentiable at x = a, f (a) = 0 = g(a), and g (a) = 0, then lim x→a f (x) f (a) = . g(x) g (a) 4.8 CONCEPTS: QUESTIONS AND DISCUSSION In the following questions, think of the given functions f and g as a tortoise and a hare racing toward infinity as x → +∞. Which is the tortoise and which is the hare? 1. f (x) = x 2 and g(x) = x 5 2. f (x) = x 1/2 and g(x) = x 1/5 ex x 10 4. f (x) = e x and g(x) is a polynomial 5. f (x) = ln x and g(x) is a polynomial 3. f (x) = x 10 ln x and g(x) = 4.8 PROBLEMS Find the limits in Problems 1 through 48. x −1 x2 − 1 2x 2 − 1 lim x→∞ 5x 2 + 3x sin x 2 lim x→0 x x −1 lim x→1 sin x ex − x − 1 lim x→0 x2 u tan u lim u→0 1 − cos u ln x lim √ x→∞ 10 x 2. lim 3. 4. x→1 5. 7. 9. 11. 13. ln(x − 9) 15. lim x→10 x − 10 e x + e−x − 2 17. lim x→0 x sin x sin 3x x→0 tan 5x x3 − 1 21. lim 2 x→1 x − 1 19. lim x + sin x 3x + cos x 2x − 1 lim x→0 3x − 1 √ x2 − 1 lim √ x→∞ 4x 2 − x ln(1 + x) lim x→0 x 2e x − x 2 − 2x − 2 lim x→0 x3 x→∞ 6. 8. 10. 12. 14. t2 + 1 16. lim t→∞ t ln t tan x 18. lim x→(π/2)− ln(cos x) 46. lim x→π/4 48. lim x→2 1 − tan x 4x − π x 5 − 5x 2 − 12 x 10 − 500x − 24 49. y = 26. 31. x→0 ln(1 + x 2 ) e x − cos x x→0 22. 25. 29. 47. lim ln(1 + x) ln(1 − x 2 ) √ 1 + 3x − 1 42. lim x→0 x √ √ 3 + 2x − 3 + x 44. lim x→0 x 40. lim 20. lim 24. 27. exp(x 3 ) − 1 x→0 x − sin x √ 3 1 + 4x − 1 43. lim x→0 x √ √ 3 1+x − 3 1−x 45. lim x→0 x 41. lim e3x − e−3x x→0 2x sec x 36. lim x→π/2 tan x 2x − sin π x 38. lim x→1/2 4x 2 − 1 34. lim Sketch the graphs of the curves in Problems 49 through 60. Even if you use a graphing calculator or computer, apply l’Hôpital’s rule as necessary to verify the apparent behavior of the curve as x approaches a point where the function has an indeterminate form. 23. lim x→∞ 300 3x − 4 2x − 5 e3x − 1 lim x→0 x √ 1 − cos x lim x→0+ x 1 − cos x lim x→0 x3 1 + cos 2z lim z→π/2 1 − sin 2z x − tan x lim x→0 x3 er lim r →∞ (r + 1)4 1. lim 2 − e x − e−x x→0 2x 2 2x − π 35. lim x→π/2 tan 2x x − 2 cos π x 37. lim x→2 x2 − 4 √ ln 2x 39. lim √ 3 x→0+ ln 3x 33. lim 28. 30. 32. e x − e−x x→0 x x3 − 8 lim x→2 x 4 − 16 √ x2 + 4 lim x→∞ x 2x lim x→∞ 3x √ x3 + x lim √ x→∞ 2x 3 − 4 ln(ln x) lim x→∞ x ln x sin x − tan x lim x→0 x3 sin2 x x sin x 51. y = x −π 1 − cos x 53. y = x2 −x 55. y = xe sin2 x x2 cos x 52. y = 2x − π x − sin x 54. y = x3 √ −x 56. y = e x 57. y = xe− ln x 59. y = x 58. y = x 2 e−2x ln x 60. y = √ √ x+ 3x √ x 61. Show that lim x→∞ 50. y = xk = 0 if k is a positive real number. ex More Indeterminate Forms SECTION 4.9 ln x = 0 if k is a positive real number. xk 63. Suppose that n is a fixed positive integer larger than 1. Show that the curve y = x n e−x has a single local maximum and two inflection points for x > 0, and also has the x-axis as an asymptote. 64. Suppose that k is an arbitrary positive real number. Show that the curve y = x −k ln x has a single local maximum and a single inflection point for x > 0, and also has the x-axis as an asymptote. 1 65. Substitute y = in Eq. (11) to show that x 62. Show that lim 301 71. Show without using l’Hôpital’s rule that x→∞ lim x k ln x = 0 lim x x e x→∞ = +∞. Thus the function f (x) = x x increases even faster than the exponential function e x as x → +∞. 72. If a chemical plant releases an amount A of a pollutant into a canal at time t = 0, then the resulting concentration of pollutant at time t in the water at a town on the canal a fixed distance x0 downstream is x→0+ if k is a positive real number. 66. Show that if n is any integer, then C(t) = √ (ln x) = 0. x 67. Suppose that f (x) is continuous. Show that A πkt exp − x02 4kt n lim x→∞ f (x + h) − f (x − h) = f (x). 2h The symmetric difference quotient on the left can be used (with h very small) to approximate the derivative numerically, and turns out to be a better approximation than the one-sided difference quotient [ f (x + h) − f (x)]/ h. 68. Suppose that f (x) is continuous. Show that where k is a constant. Sketch a typical graph of C(t) for t 0. Then show that the maximum pollutant concentration that occurs at the town is lim h→0 f (x + h) − 2 f (x) + f (x − h) = f (x). h2 The second difference quotient on the left can be used (with h very small) to approximate the second derivative numerically. 69. In his calculus textbook of 1696, l’Hôpital used a limit similar to √ √ 2x − x 4 − 3 x lim x→1 1 − x 4/3 to illustrate his rule. Evaluate this limit. 70. Establish the 0/0 version of l’Hôpital’s rule for the case a = ∞. Suggestion: Let F(t) = f (1/t) and G(t) = g(1/t). Then show that f (x) F(t) F (t) f (x) = lim = lim = lim , lim x→∞ g(x) x→∞ g (x) t→0+ G(t) t→0+ G (t) lim Cmax A = x0 2 . πe 73. (a) If f (x) = x n e−x (where n is a fixed positive integer) is the function of Problem 63, show that the maximum value of f (x) for x 0 is f (n) = n n e−n . (b) Conclude from the fact that f (n − 1) and f (n + 1) are both less than f (n) that h→0 using l’Hôpital’s rule for the case a = 0. 1+ 1 n n < e < 1− 1 n −n . Substitute n = 1,000,000 to prove that e = 2.71828 accurate to five decimal places. 74. (a) Approximate numerically the solution x1 of the equation ln x = x 1/10 that is indicated in Fig. 4.8.8. (b) Use a calculator or computer to plot the graphs of y = ln x and y = x 1/10 in a viewing window that shows a second solution x2 of the equation ln x = x 1/10 . Then approximate x2 numerically. Suggestion: Plot the graphs on successive intervals of the form [10n , 10n+1 ] where n = 1, 2, 3, . . . . By the time you locate x2 you may well have the feeling of “going boldly where no one has gone before!” 4.9 MORE INDETERMINATE FORMS We saw in Section 4.8 that l’Hôpital’s rule can be applied to the indeterminate forms 0/0 and ∞/∞. There are other indeterminate forms; although l’Hôpital’s rule cannot be applied directly to these other forms, it may be possible to convert them into the form 0/0 or into the form ∞/∞. If so, it may be possible to apply l’Hôpital’s rule. Suppose that lim f (x) = 0 x→a and lim g(x) = ∞. x→a Then we say that the product f (x) · g(x) has the indeterminate form 0 · ∞ at x = a (or as x → a). To find the limit of f (x) · g(x) at x = a, we can change the problem to 301 302 CHAPTER 4 Additional Applications of the Derivative one of the forms 0/0 or ∞/∞ in this way: f (x) · g(x) = f (x) g(x) = . 1/g(x) 1/f (x) Now l’Hôpital’s rule may be applied if its other hypotheses are satisfied, as illustrated in Example 1. EXAMPLE 1 Find lim x ln x→∞ x −1 . x +1 Solution We are dealing with the indeterminate form 0 · ∞, so we write lim x ln x→∞ x −1 x +1 ln = lim x→∞ x −1 x +1 . 1 x The right-hand limit has the form 0/0, so we can apply l’Hôpital’s rule. First we note that x −1 x +1 Dx ln 2 . x2 − 1 = Thus x −1 lim x ln x→∞ x +1 0 −0.5 x=1 −1 −1.5 y −2 −2.5 −3 y = −2 y = x ln −4 = lim x→∞ x−1 x+1 −3.5 2 2−1 x = lim 1 x→∞ − 2 x Hence the curve y = x ln 0 0.5 1 1.5 2 2.5 3 3.5 4 x FIGURE 4.9.1 Visual corroboration of the limit in Example 1. −2x 2 = lim x 2 − 1 x→∞ x −1 , x +1 −2 = −2. 1 1− 2 x x > 1, has the line y = −2 as a horizontal asymptote as x → +∞. (See Fig. 4.9.1.) It also ◗ has the line x = 1 as a vertical asymptote as x → 1+ . (Why?) If lim f (x) = +∞ = lim g(x), x→a x→a then we say that f (x) − g(x) has the indeterminate form ∞ − ∞ as x → a. To evaluate lim [ f (x) − g(x)], x→a we try by algebraic manipulation to convert f (x) − g(x) into a form of type 0/0 or ∞/∞ so that it may be possible to apply l’Hôpital’s rule. If f (x) or g(x) is expressed as a fraction, we can sometimes do this by finding a common denominator. In most cases, however, subtler methods are required. Example 2 illustrates the technique of finding a common denominator. Example 3 demonstrates a factoring technique that can be effective. 302 More Indeterminate Forms SECTION 4.9 303 EXAMPLE 2 lim x→0 1 1 − x sin x = lim (sin x) − x x sin x (form 0/0 ) = lim (cos x) − 1 sin x + x cos x (still 0/0) = lim − sin x = 0. 2 cos x − x sin x x→0 x→0 x→0 EXAMPLE 3 lim x→+∞ x2 + 3x − x = lim x x→+∞ 3 1+ −1 x ◗ (form ∞ · 0) 3 −1 x (form 0/0 now) = lim 1 x→+∞ x 3 −1/2 3 1 − x2 1+ 2 x = lim 1 x→+∞ − 2 x 3 3 = lim 2 = . x→+∞ 2 3 1+ x 1+ 3 2.5 2 1.5 1 y 0.5 0 −0.5 −1 −1.5 −2 y= y= 0 2 3 2 x2 + 3x − x 4 6 8 10 x FIGURE 4.9.2 Visual corroboration of the limit in Example 3. √ Thus the curve y = x 2 + 3x − x, x > 0, has the line y = as x → +∞. (See Fig. 4.9.2.) 3 2 as a horizontal asymptote ◗ The Indeterminate Forms 00 , ∞0 , and 1∞ Suppose that we need to find the limit of a quantity y = [ f (x)]g(x) , where the limits of f and g as x → a are such that one of the indeterminate forms 00 , ∞0 , or 1∞ is produced. We first compute the natural logarithm ln y = ln [ f (x)]g(x) = g(x) ln f (x). For each of the three indeterminate forms mentioned here, g(x) ln f (x) has the form 0 · ∞, so we can use our earlier methods to find L = limx→a ln y (assuming that f (x) > 0 near x = a, so that y > 0). Then lim [ f (x)]g(x) = lim y = lim exp(ln y) = exp lim ln y = e L , x→a x→a x→a x→a because the exponential function is continuous. Thus we have the following four steps for finding the limit of [ f (x)]g(x) as x → a: 1. Let y = [ f (x)]g(x) . 2. Simplify ln y = g(x) ln f (x). 3. Evaluate L = lim ln y. x→a 4. Conclude that lim [ f (x)]g(x) = e L . x→a 303 304 CHAPTER 4 Additional Applications of the Derivative 2 1 0.9 0.8 1 ≈ 0.607 0.7 y = e 0.6 y 0.5 0.4 0.3 2 y = (cos x)1/x 0.2 0.1 0 −1.5 −1 −0.5 0 x EXAMPLE 4 Find lim (cos x)1/x . x→0 2 Solution Here we have the indeterminate form 1∞ . If we let y = (cos x)1/x , then ln cos x 2 . ln y = ln (cos x)1/x = x2 As x → 0, cos x → 1, and so ln cos x → 0; we are now dealing with the indeterminate form 0/0. Hence two applications of l’Hôpital’s rule yield 0.5 1 1.5 FIGURE 4.9.3 Visual corroboration of the limit in Example 4. ln cos x (−sin x)/(cos x) −tan x = lim = lim x→0 x→0 x2 2x 2x −sec2 x 1 = lim =− . x→0 2 2 lim ln y = lim x→0 (0/0 form) x→0 Consequently, as suggested by Fig. 4.9.3, 1 2 lim (cos x)1/x = e−1/2 = √ . e ◗ x→0 EXAMPLE 5 Find lim x tan x . x→0+ 5 4.5 4 3.5 3 y 2.5 2 1.5 1 0.5 0 Solution This has the indeterminate form 00 . If y = x tan x , then ln x . cot x ln y = (tan x)(ln x) = Now we have the indeterminate form ∞/∞, and l’Hôpital’s rule yields y = x tan x 0 0.5 1 1.5 x 1 ln x sin2 x x = lim = − lim lim ln y = lim x x→0+ x→0+ cot x x→0+ −csc2 x x→0+ sin x (sin x) = (−1) · 0 = 0. = − lim x x→0+ Therefore, limx→0+ x tan x = e0 = 1. The graph of the curve y = x tan x , 0 < x < π/2 in Fig. 4.9.4 provides corroboration of this limit. We note also a local minimum on the curve near x = 0.4. (See Problem 45.) ◗ FIGURE 4.9.4 Visual corroboration of the limit in Example 5. Although a 0 = 1 for any nonzero constant a, the form 00 is indeterminate—the limit is not necessarily 1 (see Problem 52). But the form 0∞ is not indeterminate; its limit is zero. For example, lim x 1/x = 0. y x→0+ y=e ( y= 1+ 1 x x ) The Number e as a Limit Figure 4.9.5 shows the graph of the function (0, 1) f (x) = 1 + x FIGURE 4.9.5 The graph 1 x y = 1+ has the horizontal x asymptote y = e. x (1) for x > 0. The graph appears to begin at the point (0, 1) and to approach a horizontal asymptote as x → +∞. Note that f (x) has the indeterminate form ∞0 as x → 0+ and has the indeterminate form 1∞ as x → +∞. In each case our strategy is to calculate the limit of ln f (x): lim ln 1 + 304 1 x 1 x x = lim x · ln 1 + 1 x ln 1 + = lim 1 x 1 x . More Indeterminate Forms SECTION 4.9 305 The last limit here has the indeterminate form ∞/∞ as x → 0+ and has the indeterminate form 0/0 as x → +∞. In each case we can apply l’Hôpital’s rule to obtain 1 lim ln 1 + x x − = lim 1 1 · x 0 x 2 1 + (1/x) = = lim 1 1 x +1 − 2 x as x → 0+ , as x → +∞. Thus we find that 1+ 1 x x lim 1+ 1 x x lim x→0+ = e0 = 1 and that x→+∞ = e1 = e. (2) The last limit shows that the horizontal asymptote in Fig. 4.9.5 is the line y = e. If we write x = n (a positive integer) in Eq. (2), we obtain the famous limit e = lim n→∞ 1+ n 1 n , (3) which can be used to approximate the number e. In Problem 44 we ask you to use l’Hôpital’s rule similarly to derive the more general limit expression x n e x = lim 1 + n→∞ n (4) for the exponential function. The limit in (3) can be approximated with a very rudimentary calculator by substituting n = 2k (a power of 2) to get k 1 2k 1+ k 2 10 20 30 40 50 2.71695 5729 2.71828 0532 2.71828 1827 2.71828 1828 2.71828 1828 FIGURE 4.9.6 Approximating the number e. e = lim k→∞ 1 1+ k 2 2k = lim ν 2 where ν = 1 + (1/2k ). Then (ν 2 )2 = ν 4 , (ν 4 )2 = ν 8 , k (5) k→∞ (ν 8 )2 = ν 16 , ... , ν2 k−1 2 k = ν2 . k Therefore we should get the value [1 + (1/2k )]2 if we enter ν = 1 + (1/2k ) and then press the x2 key k times in succession. (Try this with your own calculator. Can you see how and why the process may fail if k is too large?) The entries in the table in Fig. 4.9.6 were calculated using a high-precision computer (rather than a simple calculator). They indicate that e = 2.71828 1828 accurate to nine decimal places. 4.9 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. x −1 = ∞ · 0 = 0. x +1 1 1 − = 0. lim x→0 x sin x 3 x 2 + 3x − x = . lim x→∞ 2 1 1/x 2 lim (cos x) =− . x→0 2 lim x tan x = 1. 1. lim x ln x→∞ 3. 5. 7. 9. x→0+ x −1 2. lim x ln = −2. x→∞ x +1 x 2 + 3x − x = ∞−∞ = 0. 4. lim x→∞ 2 6. lim (cos x)1/x = 1∞ = 1. x→0 2 8. lim (cos x)1/x = e−1/2 . x→0 10. lim x tan x = 00 = 0. x→0+ 305 306 CHAPTER 4 Additional Applications of the Derivative 4.9 CONCEPTS: QUESTIONS AND DISCUSSION 1. List the seven indeterminate forms discussed in Sections 4.8 and 4.9. Illustrate each of these forms with your own example. 2. Explain in your own words why 0∞ and 0−∞ are not indeterminate forms. 3. Suppose that someone claims that “calculus is merely the study of indeterminate forms.” Based on what you’ve learned so far, how would you argue for (or against) this claim? 4.9 PROBLEMS Find the limits in Problems 1 through 34. 1. lim x cot x 2. lim x→0 x→0 25. lim x→0 1 − cot x x 5. lim x 2 csc2 x 6. lim e−x ln x x→0+ 8. lim x→2 x→0+ x→0+ x→0 x→∞ 1 x − x2 + x − 2 x −1 √ √ 16. lim x +1− x 15. lim x→1+ x→∞ 1 1 − x→0 x ln(1 + x) 18. lim x2 + x − x2 − x x→∞ 3 19. lim x 3 + 2x + 5 − x 20. lim x x 17. lim x→0+ 1 x2 x 21. lim x sin x 22. lim 23. lim (ln x)1/x 24. lim 1− x→∞ 2 1.8 1.6 1.4 1.2 y 1 0.8 0.6 0.4 0.2 0 0 x→∞ 0.5 FIGURE 4.9.7 306 x 2x − 1 2x + 1 x→∞ x 1 1.5 x→1+ Figures 4.9.7 through 4.9.9 illustrate the graphs of some of the functions defined for x > 0 in Problems 35 through 42. In each of these problems: (a) First use your own calculator or computer to graph the given function f (x) with an x-range sufficient to suggest its behavior both as x → 0+ and as x → +∞. (b) Then apply l’Hôpital’s rule as necessary to verify this suspected behavior near zero and +∞. (c) Finally, estimate graphically and/or numerically the maximum value attained by f (x) for x 0. If possible, find this maximum value exactly. 1 1 √ − sin x x x→0+ x→π/2 1 1 − √ x −2 x2 − 4 5 34. lim x 5 − 3x 4 + 17 − x x→∞ x→∞ 32. lim (x − 1)ln x 30. lim (tan x − sec x) x→2+ 1 1 − x x e −1 14. lim 31. lim x 1/(1−x) x→0+ 33. lim 12. lim (x − sin x) exp(−x 2 ) x→π 13. lim 1 1 − x − 2 ln(x − 1) x→π/2 11. lim (x − π ) csc x 28. lim (sin x)sec x cos x→1 10. lim (tan x)(cos 3x) 9. lim x ln x x→0+ x→0+ x→∞ x→∞ 26. lim (1 + 2x)1/(3x) 1 x→∞ x2 29. lim (x + sin x)x 27. lim 4. lim (sin x)(ln sin x) 7. lim x(e1/x − 1) 1/x 2 x4 1 7x + 8 3. lim ln x→0 x 4x + 8 x→0 sin x x 35. f (x) = x 1/x 37. f (x) = (x 2 )1/x 36. f (x) = x (1/x 38. f (x) = x −x 2) 39. f (x) = (1 + x 2 )1/x 40. f (x) = 1 + 1 x x2 1/x 2 (cos x−1) 42. f (x) = e 41. f (x) = (x + sin x)1/x 2 1.8 1.6 1.4 1.2 y 1 0.8 0.6 0.4 0.2 0 Use l’Hôpital’s rule to establish the limits in Problems 43 and 44. x n 44. lim 1 + = ex 43. lim (1 + hx)1/ h = e x n→∞ h→0 n 3 2.5 2 y 1.5 1 0.5 0 1 2 FIGURE 4.9.8 x 3 4 5 0 0 5 10 FIGURE 4.9.9 x 15 20 25 More Indeterminate Forms SECTION 4.9 45. Estimate graphically or numerically the location of the local minimum point on the graph y = x tan x shown in Fig. 4.9.4. 46. Let n be a fixed positive integer and let p(x) be the polynomial p(x) = x n + a1 x n−1 + a2 x n−2 + · · · + an−1 x + an ; the numbers a1 , a2 , . . . , an are fixed real numbers. Prove that a1 lim [ p(x)]1/n − x = . x→∞ n 47. As we shall see in Problem 52 of Section 7.6, the surface area of the ellipsoid obtained by revolving the ellipse y2 x2 + =1 a2 b2 around the x-axis is (a > b > 0) a c b −1 , A = 2πab + sin a c a √ where c = a 2 − b2 . Use l’Hôpital’s rule to show that lim A = 4πa 2 , b→a the surface area of a sphere of radius a. 48. If the amount A 0 is invested in an account that earns interest at the annual rate r 1 compounded n times annually, then the amount A in the account after t years is given by A = A0 1 + r n 307 50. Graph the function f (x) = | ln x|1/| ln x| for x > 0 and determine its behavior as x → 0+ , as x → +∞, and as x approaches 1 from either side. 51. Graph the function f (x) = | ln x|| ln x| for x > 0 and determine its behavior as x → 0+ , as x → +∞, and as x approaches 1 from either side. Explore both graphically (by zooming) and symbolically (by differentiating) the question of whether f is differentiable at x = 1. 52. Let α be a fixed real number. (a) Evaluate (in terms of α) the 00 indeterminate form αx 2 1 . lim exp − 2 x→0 x (Note that l’Hôpital’s rule is not needed.) Thus the indeterminate form 00 may have as its limit any positive real number. Explain why. (b) Can the limit of a 00 indeterminate form be zero, negative, or infinite? Explain. 53. Sketch the graph of the function f (x) = (1 + x)1/x for x −1, x = 0. Explain why you can approximate the number e by zooming in on the apparent y-intercept of this graph. Do so, accurate to five decimal places. 54. This problem explores the fact that a lead ball hits the ground with greater speed than a feather when both are dropped simultaneously from the top of a tall building. Assuming that air resistance is proportional to downward velocity v, we will show in Chapter 8 that after t seconds the velocity of a dropped body of mass m is given by v(t) = mg 1 − e−kt/m k nt . (a) Show that A is an increasing function of n (with r and t fixed). Thus the bank that compounds more often pays more interest. (b) Use l’Hôpital’s rule to show that lim A(n) = A 0 er t . where g is the familiar acceleration of gravity and k denotes a constant air-resistance coefficient. (a) Note that lim v(t) = t→∞ Thus a body’s velocity tends to a finite limit after it has fallen a sufficiently long time. (b) Note that n→∞ This is the amount after t years if the bank compounds interest “continuously.” The “annual yield” is the value of this limit in the case t = 1. (c) If a bank advertises an annual interest rate of 8% compounded continuously, what is the annual yield? 49. Graph the function f (x) = | ln x|1/x for x > 0 and determine its behavior as x → 0+ and as x → +∞. Estimate graphically and/or numerically the locations of any critical points or inflection points on the graph of f . mg . k lim v(t) = 0. m→0 Consequently a light “feathery” body falls very slowly through the air. (c) Show that lim v(t) = gt = lim v(t), m→∞ k→0+ the velocity of the body after t seconds in the case of no air resistance. Thus a very heavy body tends to fall much as it would with no air resistance. 307 308 CHAPTER 4 Additional Applications of the Derivative CHAPTER 4: REVIEW Understanding: Concepts, Definitions, Results Refer to the listed pages to review the concepts, definitions, and formulas in this chapter that you need to understand. Section Pages 4.2 The increment y and the differential dy of a function y = f (x) . . . . . . . . . . . . . . . . . . . . 226 The linear approximation formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226 The linear approximation to f (x) near the point x = a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Absolute and relative errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 The error y − dy in linear approximation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 Differentiation rules in differential form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231–232 4.3 Increasing functions and decreasing functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 Geometric interpretation of the mean value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 Rolle’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Statement and proof of the mean value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237–238 Constant functions and zero derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 Functions with equal derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 Significance of the sign of the first derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235, 240 4.4 Distinction between local (or relative) and global (or absolute extrema) . . . . . . . . . . . . . . 246 The first derivative test for local extrema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 Open-interval maximum-minimum problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 The first derivative test for global extrema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252 4.5 Steps in graphing polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 Behavior at infinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 Critical points and increasing/decreasing behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 4.6 Second and higher derivatives of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 The significance of the sign of the second derivative: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267–268 bending downward and bending upward The second derivative test for local extrema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268 The definition of concavity on an interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 The test for concavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 The definition of inflection points and the inflection point test . . . . . . . . . . . . . . . . . . . . . . . 272 Use of inflection points in curve-sketching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 4.7 Infinite limits of function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 Vertical asymptotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Limits at infinity—that is, as x → ±∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 Horizontal asymptotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Curve-sketching strategy—putting it all together . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 Slant asymptotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 4.8 L’Hôpital’s rule and the indeterminate form 0/0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293–294 The indeterminate form ∞/∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 The order of magnitude of exponential and logarithmic functions . . . . . . . . . . . . . . . . . . . . 296–298 4.9 The indeterminate forms 0 · ∞ and ∞ − ∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301–302 The indeterminate forms 00 , ∞0 , and 1∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 The number e as a limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 308 Chapter 4 Miscellaneous Problems 309 CHAPTER 4: REVIEW (Continued) Objectives: Methods and Techniques Work the listed problems in each section to practice the methods and techniques in this chapter that you need to master. Section Problems 4.2 Calculating differentials of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 5, 9, 13 Finding linear approximations to functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17, 23 Calculating numerical linear approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25, 31, 33 Applying differentials in geometric situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41, 43, 49 4.3 Using increasing-decreasing behavior to match functions and graphs . . . . . . . . . . . . . . . . . 1, 3 Determining the increasing-decreasing intervals for a function . . . . . . . . . . . . . . . . . . . . . . 11, 13, 19, 21 Checking hypotheses and conclusions for Rolle’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 27, 31 Checking hypotheses and conclusions for the mean value theorem . . . . . . . . . . . . . . . . . . . 33, 35 4.4 Using the first derivative test to classify critical points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 7, 13, 21, 23 Solving applied open-interval optimization problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31, 33, 35, 41, 45 4.5 Using behavior at infinity to match functions and graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 3 Finding critical points and increasing-decreasing behavior . . . . . . . . . . . . . . . . . . . . . . . . . . 7, 11 Sketching graphs of given polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15, 19, 23, 27 4.6 Calculating higher derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 13, 17 Finding critical and inflection points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23, 27 Applying the second derivative and inflection point tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33, 35, 47 Using concavity and critical-inflection points to sketch graphs . . . . . . . . . . . . . . . . . . . . . . 63, 67, 75 Matching graphs of functions and of their second derivatives . . . . . . . . . . . . . . . . . . . . . . . . 77, 79 4.7 Investigating infinite limits and limits at infinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 3, 9 Using asymptotes to match functions and their graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19, 21, 25 Sketching graphs with extrema, inflection points, and asymptotes . . . . . . . . . . . . . . . . . . . . 35, 39, 43, 47, 49 4.8 Applying l’Hôpital’s rule to the forms 0/0 and ∞/∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 9, 13, 19, 25, 29, 33 4.9 Applying l’Hôpital’s rule to the forms 0 · ∞ and ∞ − ∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . 1, 7, 9, 13, 17 Applying l’Hôpital’s rule to the forms 00 , ∞0 , and 1∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21, 23, 31 MISCELLANEOUS PROBLEMS In Problems 1 through 6, write dy in terms of x and d x. √ 1. y = (4x − x 2 )3/2 2. y = 8x 3 x 2 + 9 x +1 4. y = sin x 2 3. y = x −1 √ x 5. y = x 2 cos x 6. y = sin 2x In Problems 7 through 16, estimate the indicated number by linear approximation. √ 7. 6401 (Note that 802 = 6400.) 1 8. 1.000007 9. (2.0003)10 (Note that 210 = 1024.) √ 10. 3 999 (Note that 103 = 1000.) √ √ 12. 3 62 11. 3 1005 √ 14. 5 30 13. 263/2 √ √ 16. 10 1000 15. 4 17 In Problems 17 through 22, estimate by linear approximation the change in the indicated quantity. 17. The volume V = s 3 of a cube, if its side length s is increased from 5 in. to 5.1 in. 18. The area A = πr of a circle, if its radius r is decreased from 10 cm to 9.8 cm. 2 19. The volume V = 43 πr 3 of a sphere, if its radius r is increased from 5 cm to 5.1 cm. 20. The volume V = 1000/ p in.3 of a gas, if the pressure p is decreased from 100 lb/in.2 to 99 lb/in.2 √ 21. The period of oscillation T = 2π L/32 of a pendulum, if its length L is increased from 2 ft to 2 ft 1 in. (Time T is in seconds and L is in feet.) 22. The lifetime L = 1030/E 13 of a light bulb with applied voltage E volts (V), if the voltage is increased from 110 V to 111 V. Compare your result with the exact change in the function L. If the mean value theorem applies to the function f on the interval [a, b], it ensures the existence of a solution c in the interval (a, b) of the equation f (c) = f (b) − f (a) . b−a In Problems 23 through 28, a function f and an interval [a, b] are given. Verify that the hypotheses of the mean value theorem are satisfied for f on [a, b]. Then use the given equation to find the value of the number c. 1 ; [1, 3] x 24. f (x) = x 3 + x − 4; [−2, 3] 23. f (x) = x − 309 310 CHAPTER 4 Additional Applications of the Derivative 25. f (x) = x 3 ; [−1, 2] x 5 ; [−1, 2] 27. f (x) = 11 5 26. f (x) = x 3 ; [−2, 1] √ 28. f (x) = x; [0, 4] Sketch the graphs of the functions in Problems 29 through 33. Indicate the local maxima and minima of each function and the intervals on which the function is increasing or decreasing. Show the concave structure of the graph and identify all inflection points. 29. 31. 33. 34. f (x) = x 2 − 6x + 4 30. f (x) = 2x 3 − 3x 2 − 36x √ f (x) = 3x 5 − 5x 3 + 60x 32. f (x) = (3 − x) x √ f (x) = (1 − x) 3 x Show that the equation x 5 + x = 5 has exactly one real solution. Calculate the first three derivatives of the functions in Problems 35 through 44. 35. f (x) = x 3 − 2x 1 1 37. g(t) = − t 2t + 1 36. f (x) = (x + 1)100 √ 38. h(y) = 3y − 1 39. f (t) = 2t 3/2 − 3t 4/3 1 x2 + 9 √ 3 42. f (z) = 3 z + √ 5 z 8 44. g(t) = (3 − t)3/2 t +2 t −2 √ 43. g(x) = 3 5 − 4x 41. h(t) = 40. g(x) = In Problems 45 through 52, calculate dy/d x and d 2 y/d x 2 under the assumption that y is defined implicitly as a function of x by the given equation. 45. 47. 49. 51. x 1/3 + y 1/3 = 1 √ y 5 − 4y + 1 = x x 2 + y 2 = 5x y + 5 y3 − y = x 2 y 46. 48. 50. 52. 2x 2 − 3x y + 5y 2 = 25 sin x y = x y x 5 + x y4 = 1 (x 2 − y 2 )2 = 4x y Sketch the graphs of the functions in Problems 53 through 72, indicating all critical points, inflection points, and asymptotes. Show the concave structure clearly. 53. f (x) = x 4 − 32x 55. f (x) = x 6 − 2x 4 √ 57. f (x) = x 3 4 − x 54. f (x) = 18x 2 − x 4 √ 56. f (x) = x x − 3 x −1 58. f (x) = x +2 x 60. f (x) = 2 x −x −2 x3 62. f (x) = 2 x −1 64. f (x) = x 4 − 2x 2 x2 + 1 x2 − 4 2x 2 f (x) = 2 x −x −2 f (x) = 3x 4 − 4x 3 x2 66. f (x) = x 3 − 12x f (x) = 2 x −1 f (x) = −10 + 6x 2 − x 3 x f (x) = ; note that 1 + x2 59. f (x) = 61. 63. 65. 67. 68. f (x) = − 69. f (x) = x 3 − 3x 70. f (x) = x 4 − 12x 2 1 1 71. f (x) = x 3 + x 2 − 5x + 3 72. f (x) = + 2 x x 73. The function 1 f (x) = 2 x + 2x + 2 has a maximum value, and only one. Find it. 74. You need to manufacture a cylindrical pot, without a top, with a volume of 1 ft3 . The cylindrical part of the pot is to be made of aluminum, the bottom of copper. Copper is five times as expensive as aluminum. What dimensions would minimize the total cost of the pot? 75. An open-topped rectangular box is to have a volume of 4500 cm3 . If its bottom is a rectangle whose length is twice its width, what dimensions would minimize the total area of the bottom and four sides of the box? 76. A small rectangular box must be made with a volume of 324 in.3 Its bottom is square and costs twice as much (per square inch) as its top and four sides. What dimensions would minimize the total cost of the material needed to make this box? 77. You must make a small rectangular box with a volume of 400 in.3 Its bottom is a rectangle whose length is twice its width. The bottom costs 7/ c/in.2 ; the top and four sides of the box cost 5/ c/in.2 What dimensions would minimize the cost of the box? 78. Suppose that f (x) is a cubic polynomial with exactly three distinct real zeros. Prove that the two zeros of f (x) are real and distinct. 79. Suppose that it costs 1 + (0.0003)v 3/2 dollars per mile to operate a truck at v miles per hour. If there are additional costs (such as the driver’s pay) of $10/hr, what speed would minimize the total cost of a 1000-mi trip? 80. The numbers a1 , a2 , . . . , an are fixed. Find a simple formula for the number x such that the sum of the squares of the distances of x from the n fixed numbers is as small as possible. 81. Sketch the curve y 2 = x(x − 1)(x − 2), showing that it consists of two pieces—one bounded and the other unbounded—and has two horizontal tangent lines, three vertical tangent lines, and two inflection points. [Suggestion: Note that the curve is symmetric around the x-axis. Begin by determining the intervals on which the product x(x − 1)(x − 2) is positive. Compute dy/d x and d 2 y/d x 2 by implicit differentiation. ] 82. Farmer Rogers wants to fence in a rectangular plot of area 2400 ft2 . She wants also to use additional fencing to build an internal divider fence parallel to two of the boundary sections (Fig. 4.MP.1). What is the minimum total length of fencing that this project will require? Verify that your answer yields the global minimum. (x − 1)(x + 1) (x 2 + 1)2 and that f (x) = 310 2x(x 2 − 3) . (x 2 + 1)3 FIGURE 4.MP.1 The fencing of Problem 82. Chapter 4 Miscellaneous Problems 83. Farmer Simmons wants to fence in a rectangular plot of area 1800 ft2 . He wants also to use additional fencing to build two internal divider fences, both parallel to the same two outer boundary sections (Fig. 4.MP.2). What is the minimum total length of fencing that this project will require? Verify that your answer yields the global minimum. 311 a fixed but unspecified positive integer). What is the minimum possible surface area of such a box? Verify that your answer yields the global minimum. 93. The graph of f (x) = x 1/3 (1−x)2/3 is shown in Fig. 4.MP.3. Recall from Section 4.7 that this graph has a slant asymptote with equation y = mx + b provided that lim [ f (x) − (mx + b)] = 0 x→+∞ or that lim [ f (x) − (mx + b)] = 0. x→−∞ (The values of m and b may be different in the two cases x → +∞ and x → −∞.) The graph here appears to have such an asymptote as x → +∞. Find m by evaluating FIGURE 4.MP.2 The fencing of Problem 83. lim x→+∞ 84. Farmer Taylor wants to fence in a rectangular plot of area 2250 m2 . She wants also to use additional fencing to build three internal divider fences, all parallel to the same two outer boundary sections. What is the minimum total length of fencing that this project will require? Verify that your answer yields the global minimum. Then find b by evaluating lim [ f (x) − mx]. x→+∞ Finally, find m and b for the case x → −∞. y 85. Farmer Upshaw wants to fence in a rectangular plot of area A ft2 . He wants also to use additional fencing to build n (a fixed but unspecified positive integer) internal divider fences, all parallel to the same two outer boundary sections. What is the minimum total length of fencing that this project will require? Verify that your answer yields the global minimum. 4 2 −4 88. A right triangle is formed in the first quadrant by a line segment that is tangent to the graph of y = 1/x and whose endpoints lie on the coordinate axes. Is there a maximum possible area of such a triangle? Is there a minimum? Justify your answers. 89. A rectangular box (with a top) is to have volume 288 in.3 , and its base is to be exactly three times as long as it is wide. What is the minimum possible surface area of such a box? Verify that your answer yields the global minimum. 90. A rectangular box (with a top) is to have volume 800 in.3 , and its base is to be exactly four times as long as it is wide. What is the minimum possible surface area of such a box? Verify that your answer yields the global minimum. 91. A rectangular box (with a top) is to have volume 225 cm3 , and its base is to be exactly five times as long as it is wide. What is the minimum possible surface area of such a box? Verify that your answer yields the global minimum. 92. A rectangular box (with a top) is to have volume V , and its base is to be exactly n times as long as it is wide (n is −2 2 4 x −2 86. What is the length of the shortest line segment that lies in the first quadrant with its endpoints on the coordinate axes and is also tangent to the graph of y = 1/x 2 ? Verify that your answer yields the global minimum. 87. A right triangle is formed in the first quadrant by a line segment that is tangent to the graph of y = 1/x 2 and whose endpoints lie on the coordinate axes. Is there a maximum possible area of such a triangle? Is there a minimum? Justify your answers. f (x) . x −4 FIGURE 4.MP.3 The graph of y = f (x) of Problem 93. 94. You are at the southernmost point of a circular lake of radius 1 mi. Your plan is to swim a straight course to another point on the shore of the lake, then jog to the northernmost point. You can jog twice as fast as you can swim. What route gives the minimum time required for your journey? Find the limits in Problems 95 through 109. x −2 x2 − 4 1 + cos x 97. lim x→π (x − π)2 tan t − sin t 99. lim t→0 t3 101. lim (cot x) ln(1 + x) 95. lim x→2 x→0 sin 2x x x − sin x 98. lim x→0 x3 ln(ln x) 100. lim x→∞ ln x 1/x 102. lim (e − 1) tan x 96. lim x→0 x→0+ x3 1 x2 1 104. lim − 2 103. lim − 2 x→∞ x + 2 x→0 x 1 − cos x x +3 √ 2 105. lim x −x −1− x x→∞ 107. lim (e2x − 2x)1/x 106. lim x 1/x x→∞ x→∞ 108. lim [1 − exp(−x )] 2 1/x 2 x→∞ 311 312 CHAPTER 4 Additional Applications of the Derivative 1 x 109. lim x 1 + −e [ Suggestion: Let u = 1/x, x→∞ x + and take the limit as u → 0 . ] is 110. According to Problem 53 of Section 7.6, the surface area of the ellipsoid obtained by revolving around the x-axis the ellipse with equation where c = A = 2πab a b+c b + ln a c a x a + y b b→a 2 =1 (0 < a < b) the surface area of a sphere of radius a. PHOTO CREDITS p. 225 (top left) The Royal Society of London; (bottom right) David E. Penney 312 , √ b2 − a 2 . Use l’Hôpital’s rule to show that lim A = 4πa 2 , 2 5 The Integral A rchimedes of Syracuse was the greatest mathematician of the ancient era from the fifth century B . C . to the second century A . D ., when the seeds of modern mathematics sprouted in Greek communities located mainly on the shores of the Mediterranean Sea. He was faArchimedes (287–212 B . C .) mous in his own time for mechanical inventions —the so-called Archimedean screw for pumping water, lever-and-pulley devices (“give me a place to stand and I can move the earth”), a planetarium that duplicated the motions of heavenly bodies so accurately as to show eclipses of the sun and moon, and machines of war that terrified Roman soldiers in the siege of Syracuse, during which Archimedes was killed. But it is said that for Archimedes himself these inventions were merely the “diversions of geometry at play,” and his writings are devoted to mathematical investigations. Archimedes carried out many area and volume computations that now use integral calculus—ranging from areas of circles, spheres, and segments of conic sections to volumes of cones, spheres, ellipsoids, and paraboloids. It had been proved earlier in Euclid’s Elements that the area A of a circle is proportional to the square of its radius r , so A = πr 2 for some proportionality constant π . But it was Archimedes who accurately approximated the numerical value of π , showing that it lies between the value 3 17 memorized by elemen. Euclid had tary school children and the lower bound 3 10 71 also proved that the volume V of a sphere of radius r is given by V = μr 3 (μ constant), but it was Archimedes who discovered (and proved) that μ = 4π/3. He also discovered the now-familiar volume formulas V = πr 2 h and V = 13 πr 2 h for the cylinder and the cone, respectively, of base radius r and height h. It was long suspected that Archimedes had not originally discovered his area and volume formulas by means of the limit-based arguments he used to establish them rigorously. In 1906 an Archimedean treatise entitled The Method was rediscovered virtually by accident after having been lost since ancient times. In it he described a “method of discovery” based on using infinitesimals much as they were employed during the invention and exploration of calculus in the seventeenth and eighteenth centuries. To commemorate his sphere and cylinder formulas, Archimedes requested that on his tombstone be carved a sphere inscribed in a circular cylinder. If the height of the cylinder is h = 2r , can you verify that the total surface areas A C and A S of the cylinder and sphere, and their volumes VC and VS , are related by Archimedes’ formulas A S = 23 A C and VS = 23 VC ? Thus the volumes and surface areas of the sphere and cylinder have the same 2 : 3 ratio. h r From Chapter 5 of Calculus, Early Transcendentals, Seventh Edition. C. Henry Edwards, David E. Penney. Copyright © 2008 by Pearson Education, Inc. All rights reserved. 313 314 CHAPTER 5 The Integral 5.1 INTRODUCTION y Slope m = ? y = f (x) P(x, f (x)) x x FIGURE 5.1.1 The tangent-line problem motivates differential calculus. Chapters 1 through 4 dealt with differential calculus, which is one of two closely related parts of the calculus. Differential calculus is centered on the concept of the derivative. Recall that the original motivation for the derivative was the problem of defining what it means for a straight line to be tangent to the graph of a function and calculating the slopes of such lines (Fig. 5.1.1). By contrast, the importance of the derivative stems from its applications to diverse problems that may, upon initial inspection, seem to have little connection with tangent lines. Integral calculus is based on the concept of the integral. The definition of the integral is motivated by the problem of defining and calculating the area of the region that lies between the graph of a positive-valued function f and the x-axis over a given closed interval [a, b]. The area of the region R of Fig. 5.1.2 is given by the integral of f from a to b, denoted by the symbol b f (x) d x. (1) a y y = f(x) R Area A = ? a b FIGURE 5.1.2 The area problem motivates integral calculus. x But the integral, like the derivative, is important due to its applications in many problems that may appear unrelated to its original motivation—problems involving motion and velocity, population growth, volume, arc length, surface area, and center of gravity, among others. The principal theorem of this chapter is the fundamental theorem of calculus in Section 5.6. It provides a vital connection between the operations of differentiation and integration, one that provides an effective method for computing values of integrals. It turns out that, in order to apply this theorem to evaluate the integral in (1), we need to find not the derivative of the function f (x) but rather a new function F(x) whose derivative is f (x): F (x) = f (x). (2) Thus we need to do “differentiation in reverse.” We therefore begin in Section 5.2 with an investigation of antidifferentiation. 5.2 ANTIDERIVATIVES AND INITIAL VALUE PROBLEMS The language of change is the natural language for the statement of most scientific laws and principles. For example, Newton’s law of cooling says that the rate of change of the temperature T of a body is proportional to the difference between T and the temperature of the surrounding medium (Fig. 5.2.1). That is, dT = −k(T − A), dt (1) where k is a positive constant and A, normally assumed to be constant, is the surrounding temperature. Similarly, the rate of change of a population P with constant birth and death rates is proportional to the size of the population: dP = kP dt (k constant). (2) Torricelli’s law of draining (Fig. 5.2.2) implies that the rate of change of the depth y of water in a draining tank is proportional to the square root of y; that is, dy √ = −k y dt (k constant). (3) Mathematical models of real-world situations frequently involve equations that contain derivatives of unknown functions. Such equations, including Eqs. (1) through (3), are called differential equations. 314 Antiderivatives and Initial Value Problems SECTION 5.2 315 Temperature A Volume V y Temperature T FIGURE 5.2.1 Newton’s law of cooling (Eq. (1)) describes the cooling of a hot rock in cold water. FIGURE 5.2.2 Torricelli’s law of draining (Eq. (3)) describes the draining of a cylindrical water tank. Antiderivatives The simplest kind of differential equation has the form dy = f (x), dx (4) where f is a given (known) function and the function y(x) is unknown. The process of finding a function from its derivative is the opposite of differentiation and is therefore called antidifferentiation. If we can find a function y(x) whose derivative is f (x), y (x) = f (x), then we call y(x) an antiderivative of f (x). DEFINITION Antiderivative An antiderivative of the function f is a function F such that F (x) = f (x) wherever f (x) is defined. The table in Fig. 5.2.3 shows some examples of functions, each paired with one of its antiderivatives. Figure 5.2.4 illustrates the operations of differentiation and antidifferentiation, beginning with the same function f and going in opposite directions. Figure 5.2.5 illustrates differentiation “undoing” the result of antidifferentiation—the derivative of the antiderivative of f (x) is the original function f (x). Antiderivative F(x) Antidifferentiation Function f (x) Antiderivative F(x) 1 2x x3 cos x sin 2x x x2 1 4 x 4 sin x − 12 cos 2x FIGURE 5.2.3 Some antiderivatives. Function f(x) Differentiation Derivative f '(x) FIGURE 5.2.4 Differentiation and antidifferentiation are opposites. F(x) Antidifferentiation Differentiation f (x) FIGURE 5.2.5 Differentiation undoes the result of antidifferentiation. 315 316 CHAPTER 5 The Integral EXAMPLE 1 Given the function f (x) = 3x 2 , F(x) = x 3 is an antiderivative of f (x), as are the functions √ H (x) = x 3 + π, and K (x) = x 3 − 2. G(x) = x 3 + 17, Indeed, J (x) = x 3 + C is an antiderivative of f (x) = 3x 2 for any choice of the constant C. ◗ Thus a single function has many antiderivatives, whereas a function can have only one derivative. If F(x) is an antiderivative of f (x), then so is F(x) + C for any choice of the constant C. The converse of this statement is more subtle: If F(x) is one antiderivative of f (x) on the interval I , then every antiderivative of f (x) on I is of the form F(x) + C. This follows directly from Corollary 2 of the mean value theorem in Section 4.3, according to which two functions with the same derivative on an interval differ only by a constant on that interval. Thus the graphs of any two antiderivatives F(x) + C1 and F(x) + C2 of the same function f (x) on the same interval I are “parallel” in the sense illustrated in Figs. 5.2.6 through 5.2.8. There we see that the constant C is the vertical distance between the curves y = F(x) and y = F(x) + C for each x in I . This is the geometric interpretation of Theorem 1. 8 4 4 C=2 C=4 −4 −2 y 0 C=0 C=−1 −2 −1 C = −3 0 x FIGURE 5.2.6 Graph of y = x 2 + C for various values of C. 3 2 1 2 1 2 y 0 C=0 C = −2 −4 C=−2 y = x2 + C y = sin x + C C=2 C=1 y 0 4 C=6 C=3 2 y = x3 + C −2 C = −4 −8 −2 −4 C = −6 −1 0 x 0 −1 −2 −3 1 2 FIGURE 5.2.7 Graph of y = x 3 + C for various values of C. −4 −2 0 x 2 4 FIGURE 5.2.8 Graph of y = sin x + C for various values of C. THEOREM 1 The Most General Antiderivative If F (x) = f (x) at each point of the open interval I , then every antiderivative G of f on I has the form G(x) = F(x) + C, (5) where C is a constant. Thus if F is any single antiderivative of f on the interval I , then the most general antiderivative of f on I has the form F(x) + C, as given in Eq. (5). The collection of all antiderivatives of the function f (x) is called the indefinite integral of f with respect to x and is denoted by f (x) d x. On the basis of Theorem 1, we write f (x) d x = F(x) + C, (6) where F(x) is any particular antiderivative of f (x). Therefore, f (x) d x = F(x) + C 316 if and only if F (x) = f (x). Antiderivatives and Initial Value Problems SECTION 5.2 317 The integral symbol is made like an elongated capital S. It is, in fact, a medieval S, used by Leibniz as an abbreviation for the Latin word summa (“sum”). We think of the combination . . . d x as a single symbol; we fill in the “blank” with the formula of the function whose antiderivative we seek. We may regard the differential d x as specifying the independent variable x both in the function f (x) and in its antiderivatives. EXAMPLE 2 The entries in Fig. 5.2.3 yield the indefinite integrals 1 d x = x + C, 2x d x = x 2 + C, x 3 d x = 14 x 4 + C, cos x d x = sin x + C, and sin 2x d x = − 12 cos 2x + C. You can verify each such formula by differentiating the right-hand side. Indeed, this is the surefire way to check any antidifferentiation: To verify that F(x) is an antiderivative of f (x), compute F (x) to see whether or not you obtain f (x). For instance, the differentiation Dx − 12 cos 2x + C = − 12 (−2 sin 2x) + 0 = sin 2x ◗ is sufficient to verify the fifth formula of this example. The differential d x in Eq. (6) specifies that the independent variable is x. But we can describe a specific antidifferentiation in terms of any independent variable that is convenient. For example, the indefinite integrals 3y 2 dy = y 3 + C, and 3u 2 du = u 3 + C 3t 2 dt = t 3 + C, mean exactly the same thing as 3x 2 d x = x 3 + C. Using Integral Formulas Every differentiation formula yields immediately—by “reversal” of the differentiation—a corresponding indefinite integral formula. The now-familiar derivatives of power functions and trigonometric and exponential functions yield the integral formulas stated in Theorem 2. THEOREM 2 Some Integral Formulas x k+1 xk dx = + C (if k = −1), k+1 1 cos kx d x = sin kx + C, k 1 sin kx d x = − cos kx + C, k 1 sec2 kx d x = tan kx + C, k (7) (8) (9) (10) 317 318 CHAPTER 5 The Integral 1 csc2 kx d x = − cot kx + C, k 1 sec kx tan kx d x = sec kx + C, k 1 csc kx cot kx d x = − csc kx + C, k 1 ekx d x = ekx + C. k and (11) (12) (13) (14) REMARK 1 The excluded case k = −1 in Eq. (7) corresponds to the fact that Dx [ln x] = 1/x if x > 0, so 1 d x = ln x + C x (x > 0). REMARK 2 Be sure you see why there is a minus sign in Eq. (9) but none in Eq. (8)! Recall that the operation of differentiation is linear, meaning that Dx [cF(x)] = cF (x) (where c is a constant) and Dx [F(x) ± G(x)] = F (x) ± G (x). It follows in the notation of antidifferentiation that c f (x) d x = c f (x) d x (c is a constant) and [ f (x) ± g(x)] d x = (15) f (x) d x ± g(x) d x. (16) We can summarize these two equations by saying that antidifferentiation is linear. In essence, then, we antidifferentiate a sum of functions by antidifferentiating each function individually. This is termwise (or term-by-term) antidifferentiation. Moreover, a constant coefficient in any such term is merely “carried through” the antidifferentiation. EXAMPLE 3 Find √ 4 x +3 x − 2 x 3 d x. Solution Just as in differentiation, we prepare for antidifferentiation by writing roots and reciprocals as powers with fractional or negative exponents. Thus 3 √ 4 3 x + 3x 1/2 − 4x −2 d x x + 3 x − 2 dx = x 3 1/2 [using Eqs. (15) and (16)] = x d x + 3 x d x − 4 x −2 d x x4 x −1 x 3/2 +3· 3 −4· +C 4 −1 2 √ 1 4 = x 4 + 2x x + + C. 4 x = 318 [using Eq. (7)] Antiderivatives and Initial Value Problems SECTION 5.2 319 There’s only one “ + C” because the surefire check verifies that 14 x 4 + 2x 3/2 + 4x −1 is a particular antiderivative. Hence any other antiderivative differs from this one by only ◗ a (single) constant C. EXAMPLE 4 (2 cos 3t + 5 sin 4t + 3e7t ) dt = 2 cos 3t dt + 5 sin 4t dt + 3 e7t dt = 2 13 sin 3t + 5 − 14 cos 4t + 3 17 e7t + C = 2 3 sin 3t − 54 cos 4t + 3 7t e 7 [using Eqs. (15) and (16)] [using Eqs. (8), (9), and (14)] + C. ◗ Equation (7) is the power rule “in reverse.” The generalized power rule in reverse is u k du = u k+1 +C k+1 (if k = −1), (17) where u = g(x) and du = g (x) d x. EXAMPLE 5 With u = x + 5 (so that du = d x), Eq. (17) yields 10 (x + 5) d x = u 10 du = 1 11 u 11 +C = 1 (x 11 + 5)11 + C. Note that, after substituting u = x + 5 and integrating with respect to u, our final step ◗ is to express the resulting antiderivative in terms of the original variable x. EXAMPLE 6 We want to find 20 d x. (4 − 5x)3 We plan to use Eq. (17) with u = 4 − 5x. But we must get the differential du = −5 d x into the act. The “constant-multiplier rule” of Eq. (15) permits us to do this: 20 d x = 20 (4 − 5x)−3 d x (4 − 5x)3 20 (18) (4 − 5x)−3 (−5 d x) = −5 (u = 4 − 5x, du = −5 d x) = −4 u −3 du = −4 · Thus u −2 +C −2 [Eq. (7) with k = −3]. 2 20 dx = + C. 3 (4 − 5x) (4 − 5x)2 The key step occurs in (18). There we, in effect, multiplied by the constant −5 inside the integral and compensated for that by dividing by −5 outside the integral. At the end it was necessary to replace u with 4 − 5x to express the antiderivative in terms of ◗ the original independent variable x. 319 320 CHAPTER 5 The Integral Very Simple Differential Equations The technique of antidifferentiation can often be used to solve a differential equation of the special form dy = f (x) dx (19) in which the dependent variable y does not appear on the right-hand side. To solve the differential equation in (19) is simply to find a function y(x) that satisfies Eq. (19)—a function whose derivative is the given function f (x). Hence the general solution of Eq. (19) is the indefinite integral y(x) = f (x) d x + C (20) of the function f (x). EXAMPLE 7 The general solution of the differential equation dy = 3x 2 dx is given by y(x) = 3x 2 d x = x 3 + C. ◗ A differential equation of the form in Eq. (19) may appear in conjunction with an initial condition, a condition of the form y(x0 ) = y0 . (21) This condition specifies the value y = y0 that the solution function y(x) must have at x = x0 . Once we have found the general solution in Eq. (20), we can determine the value of the constant C by substituting the information that y = y0 when x = x0 . With this specific value of C, Eq. (20) then gives the particular solution of the differential equation in (19) that satisfies the initial condition in Eq. (21). The combination dy = f (x), dx 4 y(x0 ) = y0 (22) of a differential equation with an initial condition is called an initial value problem. 0 C=2 EXAMPLE 8 Solve the initial value problem C=0 y −4 dy = 2x + 3, dx C = −2 C = −4 −8 C = −6 −4 y = x2 + 3x + C 0 4 x FIGURE 5.2.9 General solutions y = x 2 + 3x + C of the differential equation in (22) (Example 8). y(1) = 2. (23) Solution By Eq. (20) the general solution of the differential equation dy/d x = 2x +3 is given by y(x) = (2x + 3) d x = x 2 + 3x + C. Figure 5.2.9 shows the graph y = x 2 + 3x + C for various value of C. The particular solution we seek corresponds to the curve in Fig. 5.2.9 that passes through the point (1, 2), thereby satisfying the initial condition y(1) = (1)2 + 3 · (1) + C = 2. It follows that 4 + C = 2, and hence that C = −2. So the desired particular solution is given by ◗ y(x) = x 2 + 3x − 2. 320 Antiderivatives and Initial Value Problems SECTION 5.2 321 REMARK The method used in Example 8 may be described as “integrating both sides of a differential equation” with respect to x: dy d x = (2x + 3) d x; dx y(x) = x 2 + 3x + C. Rectilinear Motion x x(t) 0 Position at time t Antidifferentiation enables us, in many important cases, to analyze the motion of a particle (or “mass point”) in terms of the forces acting on it. If the particle moves in rectilinear motion along a straight line—the x-axis, for instance—under the influence of a given (possibly variable) force, then (as in Section 3.1) the motion of the particle is described by its position function x = x(t), FIGURE 5.2.10 The position function x(t) of a particle moving along the x-axis. (24) which gives its x-coordinate at time t (Fig. 5.2.10). The particle’s velocity v(t) is the time derivative of its position function, v(t) = dx , dt (25) and its acceleration a(t) is the time derivative of its velocity: a(t) = x(0) = x0 Time t = 0; velocity x'(0) = FIGURE 5.2.11 Initial data for linear motion. (26) In a typical situation, the following information is given (Fig. 5.2.11): x 0 dv d2x = 2. dt dt 0 a(t) the particle’s acceleration; x(0) = x0 its initial position; v(0) = v0 its initial velocity. (27) In principle, we can then proceed as follows to find the particle’s position function x(t). First we solve the initial value problem dv = a(t), dt v(0) = v0 (28) for the velocity function v(t). Knowing v(t), we then solve the initial value problem dx = v(t), dt x(0) = x0 (29) for the particle’s position function x(t). Thus we determine x(t) from the acceleration and initial data given in Eq. (27) by solving two successive initial value problems. For this purpose we can use the integral versions v(t) = and a(t) dt (30) v(t) dt (31) x(t) = of the derivative formulas in (25) and (26), remembering that each antidifferentiation involves an arbitrary constant. 321 322 CHAPTER 5 The Integral EXAMPLE 9 A particle starts from rest (that is, with initial velocity zero) at the point x = 10 and moves along the x-axis with acceleration function a(t) = 12t. Find its resulting position function x(t). Solution First we must solve the initial value problem dv = a(t) = 12t, v(0) = 0 dt to find the velocity function v(t). Using Eq. (30) we get v(t) = a(t) dt = 12t dt = 6t 2 + C1 . (We write C1 because we anticipate the appearance of a second constant when we integrate again to find x(t).) Then substituting the initial data t = 0, v = 0 yields 0 = 6 · 02 + C 1 = C 1 , so it follows that v(t) = 6t 2 . Next we must solve the initial value problem dx = v(t) = 6t 2 , x(0) = 10 dt for x(t). Using Eq. (31) we get x(t) = v(t) dt = 6t 2 dt = 2t 3 + C2 . Then substituting the initial data t = 0, x = 10 yields 10 = 2 · 03 + C2 = C2 , so it follows finally that the particle’s position function is x(t) = 2t 3 + 10. ◗ Constant Acceleration The solution of the initial value problems in Eqs. (28) and (29) is simplest when the given acceleration a is constant. We begin with dv =a dt (a is a constant) and antidifferentiate: v(t) = a dt. So v(t) = at + C1 . (32) To evaluate the constant C1 , we substitute the initial value v(0) = v0 ; this gives v0 = a · 0 + C1 = C1 . Therefore, Eq. (32) becomes v(t) = at + v0 . (33) Because x (t) = v(t), a second antidifferentiation yields x(t) = v(t) dt = (at + v0 ) dt; x(t) = 12 at 2 + v0 t + C2 . 322 (34) Antiderivatives and Initial Value Problems SECTION 5.2 323 Now substituting the initial value x(0) = x0 gives x0 = 12 a · (0)2 + v0 · (0) + C2 = C2 in Eq. (34). Thus the position function of the particle is x(t) = 12 at 2 + v0 t + x0 . (35) WARNING Equations (33) and (35) are valid only in the case of constant acceleration a. They do not apply to problems in which the acceleration varies. EXAMPLE 10 The skid marks made by an automobile indicate that its brakes were fully applied for a distance of 160 ft before it came to a stop. Suppose that the car in question has a constant deceleration of 20 ft/s2 under the conditions of the skid. How fast was the car traveling when its brakes were first applied? Solution The introduction of a convenient coordinate system is often crucial to the successful solution of a physical problem. Here we take the x-axis to be positively oriented in the direction of motion of the car. We choose the origin so that x0 = 0 when t = 0, the time when the brakes were first applied (Fig. 5.2.12). In this coordinate system, the car’s velocity v(t) is a decreasing function of time t (in seconds), so its acceleration is a = −20 (ft/s2 ) rather than a = +20. Hence we begin with the constant acceleration equation dv = −20. dt Constant deceleration a = −20 x Start: t = 0 x=0 = 0 Stop: x = 160 =0 FIGURE 5.2.12 Skid marks 160 ft long (Example 10). Antidifferentiation as in Eq. (30) gives v(t) = (−20) dt = −20t + C1 . Even though the initial velocity is unknown and not given, the initial data t = 0, v = v0 still yield C1 = v0 . So the car’s velocity function is v(t) = −20t + v0 . (36) A second antidifferentiation as in Eq. (31) gives x(t) = (−20t + v0 ) dt = −10t 2 + v0 t + C2 . Substituting the initial data t = 0, x0 = 0 yields C2 = 0, so the position function of the car is x(t) = −10t 2 + v0 t. (37) The fact that the skid marks are 160 ft long tells us that x = 160 when the car comes to a stop; that is, x = 160 when v = 0. 323 324 CHAPTER 5 The Integral Substituting these values into the velocity and position equations [Eqs. (36) and (37)] then yields the two simultaneous equations −20t + v0 = 0, −10t 2 + v0 t = 160. We now solve these for v0 and t to find the initial velocity v0 and the duration t of the car’s skid. If we multiply the first equation by −t and add the result to the second equation, we find that 10t 2 = 160, so t = 4 when the car first comes to a stop. It follows that the velocity of the car was v0 = 20 · 4 = 80 (ft/s), or about 55 mi/h, when the brakes were first applied. ◗ Vertical Motion with Constant Gravitational Acceleration Stroboscopic photograph of a ball falling with constant acceleration due to gravity. y Position at time t One common application of Eqs. (33) and (35) involves vertical motion near the surface of the earth. A particle in such motion is subject to a downward acceleration a, which is almost exactly constant if only small vertical distances are involved. The magnitude of this constant is denoted by g, approximately 32 ft/s2 or 9.8 m/s2 . (If you need more accurate values for g, use 32.17 ft/s2 in the fps system or 9.807 m/s2 in the mks system.) If we neglect air resistance, we may assume that this acceleration due to gravity is the only outside influence on the moving particle. Because we deal with vertical motion here, it is natural to choose the y-axis as the coordinate system for the position of the particle and to place “ground level” where y = 0 (Fig. 5.2.13). If we choose the upward direction to be the positive direction, then the effect of gravity on the particle is to decrease its height and also to decrease its velocity v = dy/dt. Then the acceleration of the particle is y(t) a= dv = −g = −32 dt (ft/s2 ). Equations (33) and (35) then become Ground y=0 FIGURE 5.2.13 The position function y(t) of a particle moving vertically. Positive values upward v(t) = −32t + v0 (38) y(t) = −16t 2 + v0 t + y0 (39) and Here y0 is the initial height of the particle in feet, v0 is its initial velocity in feet per second, and time t is measured in seconds. y EXAMPLE 11 Suppose that a bolt was fired vertically upward from a crossbow at ground level and that it struck the ground 20 s later. If air resistance may be neglected, find the initial velocity of the bolt and the maximum altitude that it reached. a(t) = −g t = 0: y(0) = y0 = 0 (0) = 0 Ground FIGURE 5.2.14 A bolt fired straight upward from a crossbow (Example 11). Solution We set up the coordinate system illustrated in Fig. 5.2.14, with ground level corresponding to y = 0, with the bolt fired at time t = 0 (in seconds), and with the positive direction being the upward direction. Units on the y-axis are in feet. We are given that y = 0 when t = 20. We lack any information about the initial velocity v0 . But we may use Eqs. (38) and (39) because we have set up a coordinate system in which the acceleration due to gravity acts in the negative direction. Thus y(t) = −16t 2 + v0 t + y0 = −16t 2 + v0 t and v(t) = −32t + v0 . 324 Antiderivatives and Initial Value Problems SECTION 5.2 325 We use the information that y = 0 when t = 20 in the first equation: 0 = −16 · 202 + 20v0 , and thus v0 = 16 · 20 = 320 (ft/s). To find the maximum altitude of the bolt, we maximize y(t) by finding the value of t for which its derivative is zero. In other words, the bolt reaches its maximum altitude when its velocity is zero: dy = −32t + v0 = 0, dt so at maximum altitude, t = v0 /32 = 10. At that time, the bolt has reached its maximum altitude of ymax = y(10) = −16 · 102 + 320 · 10 = 1600 (ft). The result seems contrary to experience. It may well suggest that air resistance cannot always be neglected, particularly not in problems involving long journeys at ◗ high velocity. 5.2 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. Torricelli’s law of draining implies that the rate of change of the depth y of water √ in a draining tank is proportional to y. 2. If F (x) = f (x), then F is called an antiderivative of f . 3. 2x d x = x 2 + C 4. If c is a constant then c f (x) d x = c 1 5. (x + 5)10 d x = (x + 5)11 + C. 11 6. k = −1 then x k d x = x k+1 + C. f (x) d x. 7. The general solution of the differential equation dy = 3x 2 dx is y(x) = x 3 + C. 8. The solution of the initial value problem dy = 2x + 3, dx y(1) = 2 is y(x) = x 2 + 3x + 2. 9. The solution of the initial value problem dv = 12t, dt v(0) = 0 is v(t) = 6t 2 . 10. If a particle moves in a straight line with velocity v(t) and constant acceleration dv a, then = a. dt 325 326 CHAPTER 5 The Integral 5.2 CONCEPTS: QUESTIONS AND DISCUSSION List corresponding features of the graphs of a function f and its antiderivative F. Hence describe a strategy whereby—given a plot showing the graphs of f and F— you can determine which is which. Apply your strategy to the following graphs, in which h is either the derivative or the antiderivative of g. 1. Figure 5.2.15 2. Figure 5.2.16 y g g y h x x h FIGURE 5.2.15 FIGURE 5.2.16 3. Figure 5.2.17 4. Figure 5.2.18 y y h h g x x g FIGURE 5.2.17 FIGURE 5.2.18 5.2 PROBLEMS Evaluate the indefinite integrals in Problems 1 through 30. 1. (3x 2 + 2x + 1) d x 2. (3t 4 + 5t − 6) dt 1 2 3 − 2 dt 4. 3. (1 − 2x + 3x ) d x t √ 3 5 3/2 5/2 x − 4 − x dx 5. + 2x − 1 d x 6. x3 x 3 1/2 2 3 t + 7 dt 8. − 2/3 d x 7. 2 x 3/4 x √ √ 4 1 3 2 2x x − √ dx dx 10. 9. x +√ 4 x x5 1 5 5 11. (4x 3 − 4x + 6) d x 12. t − 2 dt 4 t 1 dx 13. 7e x/7 d x 14. 7x 15. (x + 1)4 d x 16. (t + 1)10 dt √ 1 d x 18. z + 1 dz 17. 7 (x − 10) √ √ 3 x (1 − x)2 d x 20. x (x + 1)3 d x 19. 21. 23. 326 2x 4 − 3x 3 + 5 dx 7x 2 (9t + 11)5 dt (3x + 4)2 dx 22. √ x 1 24. dz (3z + 10)7 25. e2x + e−2x d x 26. x 2 e + e−x d x (5 cos 10x − 10 sin 5x) d x 27. (2 cos π x + 3 sin π x) d x 28. (3 cos π t + cos 3πt) dt 29. 30. (4 sin 2π t − 2 sin 4πt) dt 31. Verify by differentiation that the integral formulas sin x cos x d x = 12 sin2 x + C1 and sin x cos x d x = − 12 cos2 x + C2 are both valid. Reconcile these seemingly different results. What is the relation between the constants C1 and C2 ? 32. Show that the obviously different functions F1 (x) = 1 1−x and F2 (x) = x 1−x are both antiderivatives of f (x) = 1/(1 − x)2 . What is the relation between F1 (x) and F2 (x)? Antiderivatives and Initial Value Problems SECTION 5.2 327 33. Use the identities sin2 x = 1 − cos 2x 2 and to find the antiderivatives sin2 x d x and cos2 x = 1 + cos 2x 2 cos2 x d x. 34. (a) First explain why sec2 x d x = tan x + C. (b) Then use 2 the identity 1 + tan x = sec2 x to find the antiderivative tan2 x d x. Solve the initial value problems in Problems 35 through 46. 35. dy = 2x + 1; y(0) = 3 dx 36. dy = (x − 2)3 ; y(2) = 1 dx 37. √ dy = x ; y(4) = 0 dx 38. dy 1 = 2 ; y(1) = 5 dx x 39. 1 dy = √ ; y(2) = −1 dx x +2 40. √ dy = x + 9; y(−4) = 0 dx 41. dy 2 = 3x 3 + 2 ; y(1) = 1 dx x 42. 3 dy = x 4 − 3x + 3 ; y(1) = −1 dx x dy = (x − 1)3 ; y(0) = 2 43. dx √ dy = x + 5; y(4) = −3 44. dx 45. dy = 6e2x ; y(0) = 10 dx 46. dy 3 = ; y(1) = 7 dx x In Problems 47 through 52, a particle moves along the x-axis with the given acceleration function a(t), initial position x(0), and initial velocity v(0). Find the particle’s position function x(t). 47. 48. 49. 50. 51. 52. a(t) = 12t − 4; x(0) = 0, v(0) = −10 a(t) = 10 − 30t; x(0) = 5, v(0) = −5 a(t) = 2t 2 ; x(0) = −7, v(0) = 3 √ a(t) = 15 t ; x(0) = 5, v(0) = 7 a(t) = sin t; x(0) = 0, v(0) = 0 a(t) = 8 cos 2t; x(0) = −2, v(0) = 4 In Problems 53 through 56, a particle starts at the origin and travels along the x-axis with the velocity function v(t) whose graph is shown in Figs. 5.2.19 through 5.2.22. Sketch the graph of the resulting position function x(t) for 0 t 10. 53. Figure 5.2.19 54. Figure 5.2.20 10 9 8 7 6 v 5 4 3 2 1 0 0 10 9 8 7 6 v 5 4 3 2 1 0 0 2 4 6 8 10 2 4 6 8 10 t t FIGURE 5.2.19 Graph of the velocity function v(t) of Problem 53. FIGURE 5.2.20 Graph of the velocity function v(t) of Problem 54. 55. Figure 5.2.21 56. Figure 5.2.22 10 9 8 7 6 v 5 4 3 2 1 0 0 10 9 8 7 6 v 5 4 3 2 1 0 0 2 4 6 8 10 2 4 6 8 10 t t FIGURE 5.2.21 Graph of the velocity function v(t) of Problem 55. FIGURE 5.2.22 Graph of the velocity function v(t) of Problem 56. Problems 57 through 73 deal with vertical motion near the surface of the earth (with air resistance considered negligible). Use g = 32 ft/s2 for the magnitude of the gravitational acceleration. 57. You throw a ball straight upward from the ground with initial velocity 96 ft/s. How high does the ball rise, and how long does it remain aloft? 58. When Alex shot a marble straight upward from ground level with his slingshot, it reached a maximum height of 400 ft. What was the marble’s initial velocity? 59. Laura drops a stone into a well; it hits bottom 3 s later. How deep is the well? 60. Fran throws a rock straight upward alongside a tree (Fig. 5.2.23). The rock rises until it is even with the top of the tree and then falls back to the ground; it remains aloft for 4 s. How tall is the tree? h=? FIGURE 5.2.23 The tree of Problem 60. 327 328 CHAPTER 5 The Integral 61. Mickey throws a ball upward with an initial velocity of 48 ft/s from the top of a building 160 ft high. The ball soon falls to the ground at the base of the building (Fig. 5.2.24). How long does the ball remain aloft, and with what speed does it strike the ground? 71. 72. y = 160 73. 74. y=0 75. FIGURE 5.2.24 The building of Problem 61. 76. 62. A ball is dropped from the top of a building 576 ft high. With what velocity should a second ball be thrown straight downward 3 s later so that the two balls hit the ground simultaneously? 63. A ball is dropped from near the top of the Empire State Building, at a height 960 ft above 34th Street. How long does it take for the ball to reach the street, and with what velocity does it strike the street? 64. Lynda shoots an arrow straight upward from the ground with initial velocity 320 ft/s. (a) How high is the arrow after exactly 3 s have elapsed? (b) At what time is the arrow exactly 1200 ft above the ground? (c) How many seconds after its release does the arrow strike the ground? 65. Bill throws a stone upward from the ground. The stone reaches a maximum height of 225 ft. What was its initial velocity? 66. Sydney drops a rock into a well in which the water surface is 98 m below the ground. How long does it take the rock to reach the water surface? How fast is the rock moving as it penetrates the water surface? 67. Gloria drops a tennis ball from the top of a building 400 ft high. How long does it take the ball to reach the ground? With what velocity does it strike the ground? 68. Kosmo throws a baseball straight downward from the top of a tall building. The initial speed of the ball is 25 ft/s. It hits the ground with a speed of 153 ft/s. How tall is the building? 69. A ball is thrown straight upward from ground level with an initial speed of 160 ft/s. What is the maximum height that the ball attains? 70. Carolyn drops a sandbag from the top of a tall building h feet high. At the same time Jon throws a ball upward from ground level from a point directly below the sandbag. With what (initial) velocity should the ball be thrown so that it 328 meets the sandbag at the halfway point, where both have altitude h/2? Kelly throws a baseball straight downward with an initial speed of 40 ft/s from the top of the Washington Monument, 555 ft high. How long does it take the baseball to reach the ground, and with what speed does it strike the ground? A rock is dropped from an initial height of h feet above the surface of the earth. Show √ that the speed with which the rock strikes the surface is 2gh . A bomb is dropped from a balloon hovering at an altitude of 800 ft. From directly below the balloon, a projectile is fired straight upward toward the bomb exactly 2 s after the bomb is released. With what initial speed should the projectile be fired in order to hit the bomb at an altitude of exactly 400 ft? A car’s brakes are applied when the car is moving at 60 mi/h (exactly 88 ft/s). The brakes provide a constant deceleration of 40 ft/s2 . How far does the car travel before coming to a stop? A car traveling at 60 mi/h (exactly 88 ft/s) skids for 176 ft after its brakes are applied. The deceleration provided by the braking system is constant. What is its value? A spacecraft is in free fall toward the surface of the moon at a speed of 1000 mi/h. Its retrorockets, when fired, provide a deceleration of 20000 mi/h2 . At what height above the surface should the astronauts activate the retrorockets to ensure a “soft touchdown” (v = 0 at impact)? (See Fig. 5.2.25.) Ignore the effect of the moon’s gravitational field. a v Lunar surface FIGURE 5.2.25 The spacecraft of Problem 76. 77. (a) What initial velocity v0 must you use to throw a ball to a maximum height of 144 ft? (b) Now suppose that you throw a ball upward with the same initial velocity v0 on the moon, where the surface gravitational acceleration is only 5.2 ft/s2 . How high will it go, and how long will it remain aloft? 78. Arthur C. Clarke’s The Wind from the Sun (1963) describes Diana, a spacecraft propelled by the solar wind. Its 2mi2 aluminized sail provides it with an acceleration of (0.001)g = 0.032 ft/s2 . If the Diana starts from rest and travels in a straight line, calculate its distance x traveled (in miles) and its velocity v (in mi/h) after 1 min, 1 h, and 1 day. 79. A driver involved in an accident claims he was going only 25 mi/h. When police tested his car, they found that when the brakes were applied at 25 mi/h, the car skidded only 45 ft before coming to a stop. The driver’s skid marks at the accident scene measured 210 ft. Assuming the same (constant) deceleration, calculate the speed at which he was traveling prior to the accident. Elementary Area Computations SECTION 5.3 329 5.3 ELEMENTARY AREA COMPUTATIONS The indefinite integrals of Section 5.2 stem from the concept of antidifferentiation. The most fundamental type of integral is the one mentioned in Section 5.1, associated with the concept of area. It is called the definite integral, or simply the integral. Surprisingly, the quite different concepts of area and antidifferentiation have a close and deep relationship. This fact, discovered and exploited by Newton and Leibniz late in the seventeenth century, is the reason the same word, integral, is used in both contexts. The Concept of Area Perhaps everyone’s first contact with the concept of area is the formula A = bh, which gives the area A of a rectangle as the product of its base length b and its height h. We next learn that the area of a triangle is half the product of its base and height. This follows because any triangle can be split into two right triangles, and every right triangle is exactly half a rectangle (Fig. 5.3.1). h b FIGURE 5.3.1 The formula for the area of a triangle, A = 12 bh, follows with the aid of this figure. FIGURE 5.3.2 Every polygon can be represented as a union of nonoverlapping triangles. Given the formula A = 12 bh for the area of a triangle, we can—in principle—find the area of any polygonal figure (a plane region bounded by a closed “curve” consisting of a finite number of straight line segments). The reason is that any polygonal figure can be divided into nonoverlapping triangles (Fig. 5.3.2), and the area of the polygonal figure is then the sum of the areas of these triangles. This approach to area dates back several thousand years to the ancient civilizations of Egypt and Babylonia. The ancient Greeks began the investigation of areas of curvilinear figures in the fourth and fifth centuries B . C . Given a plane region R whose area they sought, they worked both with a polygonal P inscribed in R (Fig. 5.3.3) and with a polygonal Q circumscribed about R (Fig. 5.3.4). If the polygons P and Q have sufficiently many sides, all short, then it would appear that their areas a(P) and a(Q) closely approximate the area of the region R. Moreover, error control is possible: We see that a(P) < a(R) < a(Q) (1) because R contains the polygon P but is contained in the polygon Q. R R P (a) P Q (b) FIGURE 5.3.3 (a) A six-sided polygon P inscribed in R; (b) a many-sided inscribed polygon P more closely approximating the area of R. Q R R (a) (b) FIGURE 5.3.4 (a) A six-sided polygon Q circumscribed around R; (b) a many-sided circumscribed polygon Q more closely approximating the area of R. 329 330 CHAPTER 5 The Integral The inequalities in (1) bracket the desired area a(R). Suppose, for instance, that calculations based on triangular dissections (as in Fig. 5.3.2) yield a(P) = 7.341 and a(Q) = 7.343. Then the resulting inequality, 7.341 < a(R) < 7.343, implies that a(R) ≈ 7.34, accurate to two decimal places. Our primary objective here is to describe a systematic technique by which to approximate the area of an appropriate curvilinear region using easily calculated polygonal areas. Areas Under Graphs We consider the type of region that is determined by a continuous positive-valued function f defined on a closed interval [a, b]. Suppose that we want to calculate the area A of the region R that lies below the curve y = f (x) and above the interval [a, b] on the x-axis (Fig. 5.3.5). The region R is bounded on the left by the vertical line x = a and on the right by the vertical line x = b. y = f(x) y = f(x) Area A of region R a b x FIGURE 5.3.5 The area under the graph of y = f (x) from x = a to x = b. a b x FIGURE 5.3.6 Vertical strips determined by a division of [a, b] into equal-length subintervals. We divide the base interval [a, b] into subintervals, all with the same length. Above each subinterval lies a vertical strip (Fig. 5.3.6), and the area of A is the sum of the areas of these strips. On each of these base subintervals, we erect a rectangle that approximates the corresponding vertical strip. We may choose either an “inscribed” or a “circumscribed” rectangle (both possibilities are illustrated in Fig. 5.3.6), or even a rectangle that is intermediate between the two. These rectangles then make up a polygon that approximates the region R, and therefore the sum of the areas of these rectangles approximates the desired area A. For example, suppose that we want to approximate the area A of the region R that lies below the parabola y = x 2 above the interval [0, 3] on the x-axis. The computer plots in Fig. 5.3.7 show successively • • • • 5 inscribed and 5 circumscribed rectangles; 10 inscribed and 10 circumscribed rectangles; 20 inscribed and 20 circumscribed rectangles; 40 inscribed and 40 circumscribed rectangles. Each collection of inscribed rectangles gives an underestimate of A, whereas each collection of circumscribed rectangles gives an overestimate of A. The “curvilinear triangles” (by which the rectangular polygons in Fig. 5.3.7 undershoot or overshoot the region R) constitute the errors in these estimates. The more rectangles we use, the more accurate the approximation. Thus, to approximate accurately the area of such a region R, we need an effective way to calculate and sum the areas of collections of rectangles like those in Fig. 5.3.7. EXAMPLE 1 As in Fig. 5.3.7, let R denote the region that lies below the graph of f (x) = x 2 and above the interval [0, 3]. Calculate the underestimate and the overestimate of the area A of R by using 5 rectangles each of width 35 . Then repeat the 3 . computations using 10 rectangles each of width 10 330 Elementary Area Computations SECTION 5.3 Inscribed Circumscribed y (a) y 8 8 4 4 1 3 x 2 y (b) 8 8 4 4 3 x 2 y 2 3 x 1 2 3 x 1 2 3 x 1 2 3 x y 8 8 4 4 1 3 x 2 y (d) 1 y 1 (c) 331 y 8 8 4 4 1 3 x 2 FIGURE 5.3.7 (a) Five inscribed and circumscribed polygons; (b) ten inscribed and circumscribed polygons; (c) twenty inscribed and circumscribed polygons; (d) forty inscribed and circumscribed polygons. 0 3 5 6 5 9 5 12 5 FIGURE 5.3.8 Five subintervals, each of length 35 (Example 1). 3 Solution First suppose that n = 5 rectangles are used. Let A 5 denote the underestimate and A5 the overestimate obtained by using 5 rectangles based on the 5 subintervals of length 35 (Fig. 5.3.8). From Fig. 5.3.7(a) we see that the heights of the 5 inscribed rectangles (the first of which is degenerate—its height is zero) are the values . Because the of the function f (x) = x 2 at the 5 left-hand endpoints 0, 35 , 65 , 95 , and 12 5 3 base of each rectangle has length 5 , we see also that A5 = 3 5 = 3 5 2 2 2 2 · (0)2 + 35 + 65 + 95 + 12 5 9 36 81 144 · 0 + 25 + 25 + 25 + 25 = 6.48. The heights of the 5 circumscribed rectangles are the values of f (x) = x 2 at the 5 331 332 CHAPTER 5 The Integral right-hand endpoints 35 , 65 , 95 , A5 = 3 5 · = 3 5 · 12 , 5 and 3, so the corresponding overestimate is 2 3 5 9 + 25 6 2 + 5 + 36 25 + 81 25 9 2 + 5 + 144 25 + 12 2 5 225 25 + 15 2 5 = 11.88. These are crude approximations to the actual area A. On the basis of this information alone, our best estimate of A might well be the average of the under- and overestimates: 6.48 + 11.88 A 5 +A 5 = = 9.18. 2 2 0 3 10 6 10 9 10 12 10 15 10 18 10 21 10 24 10 27 10 FIGURE 5.3.9 Ten subintervals, 3 each of length 10 (Example 1). 3 Let us see if doubling the number of subintervals to n = 10 increases the accuracy significantly. Looking at Fig. 5.3.7(b), we see that the heights of the 10 inscribed 3 6 9 12 , 10 , 10 , 10 , rectangles are the values of f (x) = x 2 at the 10 left-hand endpoints 0, 10 15 18 21 24 27 , , , , and of the subintervals in Fig. 5.3.9. The base of each rectangle has 10 10 10 10 10 3 length 10 , so the resulting underestimate is A 10 = 3 10 + 3 2 6 2 9 2 12 2 · (0)2 + 10 + 10 + 10 + 10 15 2 10 + · 0+ = 3 10 = 7695 1000 18 2 10 9 100 + + 36 100 21 2 10 + + 81 100 24 2 + 10 144 100 + + 27 2 10 225 100 + 324 100 + 441 100 + 576 100 + 729 100 = 7.695. Similarly, the sum of the areas of the 10 circumscribed rectangles in Fig. 5.3.7(b) is the overestimate 2 2 2 2 2 3 3 6 9 A 10 = 10 · 10 + 10 + 10 + 12 + 15 10 10 + = 18 2 10 10395 1000 + 21 2 10 + 24 2 10 + 27 2 10 + 30 2 10 = 10.395. At this point, our best estimate of the actual area A might be the average 7.695 + 10.395 A 10 +A 10 = = 9.045. 2 2 ◗ We used a computer to calculate more refined underestimates and overestimates of the area A under the graph y = x 2 over [0, 3], with 20, 40, 80, 160, and, finally, 320 rectangles. The results (rounded to four decimal places) are shown in Fig. 5.3.10. The average values in the final column of the table suggest that A ≈ 9. Number of rectangles Underestimate Overestimate Average 5 10 20 40 80 160 320 6.4800 7.6950 8.3363 8.6653 8.8320 8.9158 8.9579 11.8800 10.3950 9.6863 9.3403 9.1695 9.0846 9.0422 9.1800 9.0450 9.0113 9.0028 9.0007 9.0002 9.0000 FIGURE 5.3.10 Estimate of the area under y = x 2 over [0, 3]. 332 Elementary Area Computations SECTION 5.3 333 Summation Notation For more convenient computation of area estimates, as in Example 1, we need a concise n ai is used to abbreviate the sum notation for sums of many numbers. The symbol i=1 of the n numbers a1 , a2 , a3 , . . . , an : n ai = a1 + a2 + a3 + · · · + an . (2) i=1 The symbol on the left here— is the capital Greek letter sigma (for S, for “sum”)— specifies the sum of the terms ai as the summation index i takes on the successive integer values from 1 to n. For instance, the sum of the squares of the first 10 positive integers is 10 i 2 = 12 + 22 + 32 + 42 + 52 + 62 + 72 + 82 + 92 + 102 i=1 = 1 + 4 + 9 + 16 + 25 + 36 + 49 + 64 + 81 + 100 = 385. The particular symbol used for the summation index is immaterial: 10 10 i2 = i=1 10 k2 = r 2 = 385. r =1 k=1 EXAMPLE 2 7 (k + 1) = 2 + 3 + 4 + 5 + 6 + 7 + 8 = 35, k=1 6 2n = 2 + 4 + 8 + 16 + 32 + 64 = 126, n=1 and 5 j=1 (−1) j+1 =1− j2 1 4 + 1 9 − 1 16 + = 1 25 3019 3600 ≈ 0.8386. ◗ The simple rules of summation n n cai = c i=1 ai (3) i=1 and n n (ai + bi ) = i=1 n ai + bi i=1 (4) i=1 are easy to verify by writing out each sum in full. Note that if ai = a (a constant) for i = 1, 2, . . . , n, then Eq. (4) yields n n n n + (a + bi ) = a+ bi = a + a + · · · + a bi , i=1 i=1 i=1 n terms i=1 and hence n n (a + bi ) = na + i=1 bi . (5) i=1 333 334 CHAPTER 5 The Integral In particular, n 1 = n. (6) i=1 The sum of the kth powers of the first n positive integers, n i k = 1k + 2k + 3k + · · · + n k , i=1 commonly occurs in area computations. The values of this sum for k = 1, 2, and 3 are given by the following formulas (see Problems 43 and 44): n(n + 1) = 12 n 2 + 12 n, 2 (7) n(n + 1)(2n + 1) = 13 n 3 + 12 n 2 + 16 n, 6 (8) n 2 (n + 1)2 = 14 n 4 + 12 n 3 + 14 n 2 . 4 (9) n i= i=1 n i2 = i=1 n i3 = i=1 EXAMPLE 3 The sum of the first 10 positive integers is given by Eq. (7) with n = 10: 10 i= 1 + 2 + 3 + · · · + 10 = i=1 10 · 11 = 55. 2 The sum of their squares and cubes are given by Eqs. (8) and (9): i2 = 10 · 11 · 21 = 385 6 i3 = 102 · 112 = 3025. 4 10 12 + 22 + 32 + · · · + 102 = i=1 and 10 13 + 23 + 33 + · · · + 103 = i=1 ◗ EXAMPLE 4 Consider the sum 10 (7i 2 − 5i) = 2 + 18 + 48 + · · · + 522 + 650. i=1 Using the rules in Eqs. (3) and (4) as well as Eqs. (7) and (8), we find that 10 10 (7i 2 − 5i) = 7 i=1 i=1 =7· 334 10 i2 − 5 i i=1 10 · 11 10 · 11 · 21 − 5· = 2420. 6 2 ◗ Elementary Area Computations SECTION 5.3 335 EXAMPLE 5 We can use Eq. (8) to simplify the evaluation of the sum for A 10 in Example 1, as follows: A 10 = 3 10 · 2 0 + 3 2 + 6 2 + ··· + 27 2 9 = 3 10 3 3 3 2 2 · 1 + 22 + 32 + · · · + 92 = 10 10 9 10 10 10 10 3 2 10 i2 i=0 = 3 10 · i2 i=1 7695 27 9 · 10 · 19 · = = 7.695. = 1000 6 1000 ◗ EXAMPLE 6 Evaluate the limit 1 + 2 + 3 + ··· + n . n2 lim n→+∞ Solution Using Eq. (7), we obtain lim n→+∞ + 1) n2 1 1 1 n+1 = lim + = , = lim n→+∞ n→+∞ 2 2n 2n 2 1 + 2 + 3 + ··· + n = lim n→+∞ n2 1 n(n 2 because the term 1/(2n) has limit zero as n → +∞. ◗ Area Sums Figure 5.3.11 shows the region R that lies below the graph of the positive-valued increasing function f and above the interval [a, b]. To approximate the area A of R, we have chosen a fixed integer n and divided the interval [a, b] into n subintervals [x0 , x1 ], [x1 , x2 ], [x2 , x3 ], ... , [xn−1 , xn ], all with the same length x = b−a . n (10) On each of the subintervals we have erected one inscribed rectangle and one circumscribed rectangle. y y = f(x) f(b) − f (a) Δx a = x0 x1 x2 xn − 1 xn = b x3 x FIGURE 5.3.11 The area under y = f (x) over the interval [a, b]. As indicated in Fig. 5.3.12, the inscribed rectangle over the ith subinterval [xi−1 , xi ] has height f (xi−1 ), whereas the ith circumscribed rectangle has height f (xi ). Because the base of each rectangle has length x, the areas of the rectangles are f (xi−1 ) x and f (xi ) x, (11) 335 336 CHAPTER 5 The Integral y y = f(x) f(xi ) f(xi − 1) Δx xi − 1 xi a = x0 xn = b x FIGURE 5.3.12 Inscribed and circumscribed rectangles on the ith subinterval [xi−1 , xi ]. respectively. Adding the areas of the inscribed rectangles for i = 1, 2, 3, . . . , n, we get the underestimate n An = f (xi−1 ) x (12) i=1 of the actual area A. Similarly, the sum of the areas of the circumscribed rectangles is the overestimate n An = f (xi ) x. (13) i=1 The inequality A n A A n then yields n n f (xi−1 ) x A i=1 f (xi ) x. (14) i=1 The inequalities in (14) would be reversed if f (x) were decreasing (rather than increasing) on [a, b]. (Why?) Areas as Limits An illustration such as Fig. 5.3.7 suggests that if the number n of subintervals is very large, so that x is small, then the areas A n andA n of the inscribed and circumscribed polygons will differ by very little. Hence both will be very close to the actual area A of the region R. We can also see this because, if f either is increasing or is decreasing on the whole interval [a, b], then the small rectangles in Fig. 5.3.11 (representing the difference between A n and A n ) can be reassembled in a “stack,” as indicated on the right in the figure. It follows that |A n − A n | = | f (b) − f (a)| x. (15) But x = (b − a)/n → 0 as n → ∞. Thus the difference between the lefthand and right-hand sums in (14) is approaching zero as n → ∞, whereas A does not change as n → ∞. It follows that the area of the region R is given by n A = lim n→∞ n f (xi−1 ) x = lim n→∞ i=1 f (xi ) x. (16) i=1 The meaning of these limits is simply that A can be found with any desired accuracy by calculating either sum in Eq. (16) with a sufficiently large number n of subintervals. In applying Eq. (16), recall that x = 336 b−a n (17) Elementary Area Computations SECTION 5.3 337 Also note that xi = a + i x (18) for i = 0, 1, 2, . . . , n, because xi is i “steps” of length x to the right of x0 = a. EXAMPLE 7 We can now compute exactly the area we approximated in Example 1—the area of the region under the graph of f (x) = x 2 over the interval [0, 3]. If we divide [0, 3] into n subintervals all of the same length, then Eqs. (17) and (18) give 3 3i 3 x = and xi = 0 + i · = n n n for i = 0, 1, 2, . . . , n. Therefore, n n n 27 n 2 3i 2 3 = 3 f (xi ) x = (xi )2 x = i . n n n i=1 i=1 i=1 i=1 Then Eq. (8) for i 2 yields n f (xi ) x = i=1 27 n3 1 1 1 1 3 1 2 1 n + n + n = 27 + + 2 . 3 2 6 3 2n 6n When we take the limit as n → ∞, Eq. (16) gives 1 1 1 + + = 9, A = lim 27 n→∞ 3 2n 6n 2 because the terms 1/(2n) and 1/(6n 2 ) approach zero as n → ∞. Thus our earlier inference from the data in Fig. 5.3.10 was correct: A = 9 exactly. ◗ EXAMPLE 8 Find the area under the graph of f (x) = 100 − 3x 2 from x = 1 to x = 5. y 80 Solution As shown in Fig. 5.3.13, the sum f (xi ) x gives the area of the inscribed rectangular polygon. With a = 1 and b = 5, Eqs. (17) and (18) give 40 1 2 3 4 FIGURE 5.3.13 The region of Example 8. x = 5 x Therefore n i=1 4 n and 4i 4 =1+ . n n 4i 2 4 f (xi ) x = 100 − 3 · 1 + n n i=1 n 48i 2 24i 4 − 2 97 − = n n n i=1 n xi = 1 + i · n 96 n 192 n 2 i − i n 2 i=1 n 3 i=1 i=1 388 96 1 2 1 192 1 3 1 2 1 = ·n− 2 n + n − 3 n + n + n n n 2 2 n 3 2 6 144 32 − 2. = 276 − n n [We have applied Eqs. (6) through (8).] Consequently, the second limit in Eq. (16) yields 144 32 − = 276 A = lim 276 − n→∞ n n = for the desired area. 388 n 1− ◗ 337 338 CHAPTER 5 The Integral Historical Note—The Number π 1 Mathematicians of ancient times tended to employ inscribed and circumscribed triangles rather than rectangles for area approximations. In the third century B . C ., Archimedes, the greatest mathematician of antiquity, used such an approach to derive the famous estimate αn Pn 223 71 αn 1 Qn FIGURE 5.3.14 Estimating π by using inscribed and circumscribed regular polygons and the unit circle. n a(Pn ) a(Q n ) 6 12 24 48 96 180 360 720 1440 2880 5760 2.598076 3.000000 3.105829 3.132629 3.139350 3.140955 3.141433 3.141553 3.141583 3.141590 3.141592 3.464102 3.215390 3.159660 3.146086 3.142715 3.141912 3.141672 3.141613 3.141598 3.141594 3.141593 = 3 10 < π < 3 17 = 71 22 . 7 Because the area of a circle of radius r is πr 2 , the number π may be defined to be the area of the unit circle of radius r = 1. We will approximate π , then, by approximating the area of the unit circle. Let Pn and Q n be n-sided regular polygons, with Pn inscribed in the unit circle and Q n circumscribed around it (Fig. 5.3.14). Because both polygons are regular, all their sides and angles are equal, so we need to find the area of only one of the triangles that we’ve shown making up Pn and one of those making up Q n . Let αn be the central angle subtended by half of one of the polygon’s sides. The angle αn is the same whether we work with Pn or with Q n . In degrees, αn = 360◦ 180◦ = . 2n n We can read various dimensions and proportions from Fig. 5.3.14. For example, we see that the area a(Pn ) = A n of Pn is given by 1 n n 360◦ (19) A n = a(Pn ) = n · 2 · sin αn cos αn = sin 2αn = sin 2 2 2 n and that the area of Q n is 1 180◦ A n = a(Q n ) = n · 2 · tan αn = n tan . 2 n (20) We substituted selected values of n into Eqs. (19) and (20) to obtain the entries of the table in Fig. 5.3.15. Because A n π A n for all n, we see that π ≈ 3.14159 to five decimal places. Archimedes’ reasoning was not circular—he used a direct method for computing the sines and cosines in Eqs. (19) and (20) that does not depend upon a priori knowledge of the value of π .∗ FIGURE 5.3.15 Data for estimating π (rounded to six-place accuracy). 5.3 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. At the beginning of Section 5.3 it is shown that the area of a triangle is half the product of its base and height. 2. In Example 1, evidence is provided to suggest that the area between the graph of f (x) = x 2 and the x-axis for 0 x 3 is 9. 10 5 i 2 = 255. 3. 4. i=1 n j=1 n 1 = n. 5. i=1 (−1) j+1 3019 = . 2 j 3600 i= 6. i=1 n(n + 1) . 2 1 + 2 + 3 + ··· + n 7. lim = 1. n→∞ n2 8. In Example 7 a proof is provided that the area of the region of Example 1 is exactly 9. ∗ 338 See Chapter 2 of C. H. Edwards, Jr., The Historical Development of the Calculus (New York: SpringerVerlag, 1979). Elementary Area Computations SECTION 5.3 339 22 223 <π < . 71 7 10. In Section 5.3 it is asserted that the area of any bounded plane region can be approximated to any degree of accuracy by the method of inscribed and circumscribed rectangles. 9. Archimedes established that 5.3 CONCEPTS: QUESTIONS AND DISCUSSION 1. When the German mathematician Carl Friedrich Gauss (1777–1855) was ten years old, his teacher asked the class to find the sum of the integers 1 through 100. Young Gauss almost immediately wrote the answer (and nothing else) on his slate. Apparently he had simply noted that the sum of the first and last of these integers is 101, as is the sum of the second and next-to-last, as is the sum of the third and second-from-last, and so forth. So he simply multiplied 101 by 50—the number of such pairs—to get the sum 5050. Explain carefully how this approach can be used to show that n i= i=1 n(n + 1) . 2 In what way does the argument depend on whether n is even or odd? 2. The area A of the region under the curve y = f (x) in Fig. 5.3.11 satisfies (for every positive integer n) the inequality A n < A < A n , where A n andA n are the sums defined in Eqs. (12) and (13). Moreover, b−a n by Eq. (15). Explain carefully why it follows—as asserted in Eq. (16)—that |A n − A n | = | f (b) − f (a)| · lim A n = lim A n = A. n→∞ n→∞ That is, explain why it follows that, given > 0, there exists an integer N such that both A n andA n differ from A by less than if n > N . 5.3 PROBLEMS Write each of the sums in Problems 1 through 8 in expanded notation. 5 6 3i 1. 2. i=1 5 3. j=1 6 5. k=1 i=1 6 1 j +1 4. 1 k2 6. (2 j − 1) j=1 6 k=1 5 7. √ 2i (−1)k+1 k2 x (−1) 8. n=1 11. 1 + 12. 1 + 13. 14. 1 2 1 3 + − 1 2 1 4 + 1 4 1 9 + + + 1 3 1 9 + + 1 + 15 4 1 + 251 16 1 1 1 1 + 16 + 32 + 64 8 1 1 1 − 81 + 243 27 4 9 √ + 8 27 8 (4i − 3) x j=1 10 6 (3i 2 + 1) 21. k=1 8 5 (r − 1)(r + 2) i=1 6 10 (i 3 − i 2 ) k=1 100 100 i2 i=1 (2k − 1)2 26. i=1 27. (i 3 − 3i + 2) 24. r =1 25. (2k − 3k 2 ) 22. i=1 23. (5 − 2 j) 20. i=1 n+1 2n−1 Write the sums in Problems 9 through 18 in summation notation. 10. 1 − 2 + 3 − 4 + 5 − 6 + 10 n=1 9. 1 + 4 + 9 + 16 + 25 2 3 19. 5 n 32 + 16 + 243 81 √ √ √ √ √ 16. 1 + 2 + 3 + 2 + 5 + 6 + 7 + 2 2 + 3 x3 x 10 x2 + + ··· + 17. x + 2 3 10 x5 x7 x 19 x3 + − + ··· − 18. x − 3 5 7 19 Use Eqs. (6) through (9) to find the sums in Problems 19 through 28. 15. i3 28. i=1 339 340 CHAPTER 5 The Integral Use the method of Example 6 to evaluate the limits in Problems 29 and 30. 43. Derive Eq. (7) by adding the equations n i = 1 + 2 + 3 + ··· + n 12 + 22 + 32 + · · · + n 2 29. lim n→∞ n3 3 3 1 + 2 + 33 + · · · + n 3 30. lim n→∞ n4 Use Eqs. (6) through (9) to derive concise formulas in terms of n for the sums in Problems 31 and 32. n n (2i − 1) 31. (2i − 1)2 32. i=1 and n i = n + (n − 1) + (n − 2) · · · + 2 + 1. i=1 44. Write the n equations obtained by substituting the values k = 1, 2, 3, . . . , n into the identity i=1 In Problems 33 through 42, let R denote the region that lies below the graph of y = f (x) over the interval [a, b] on the x-axis. Use the method of Example 1 to calculate both an underestimate An and an overestimate A n for the area A of R, based on a division of [a, b] into n subintervals all with the same length x = (b − a)/n. 33. 34. 35. 36. i=1 f (x) = x on [0, 1]; n = 5 f (x) = x on [1, 3]; n = 5 f (x) = 2x + 3 on [0, 3]; n = 6 f (x) = 13 − 3x on [0, 3]; n = 6 (Fig. 5.3.16) (k + 1)3 − k 3 = 3k 2 + 3k + 1. Add these n equations and use their sum to deduce Eq. (8) from Eq. (7). In Problems 45 through 50, first calculate (in terms of n) the sum n f (xi ) x i=1 to approximate the area A of the region under y = f (x) above the interval [a, b]. Then find A exactly (as in Examples 7 and 8) by taking the limit as n → ∞. 45. 47. 49. 51. y 12 8 4 1 3 x 2 f (x) = x on [0, 1] 46. f (x) = x 2 on [0, 2] 48. f (x) = x + 2 on [0, 2] f (x) = x 3 on [0, 3] f (x) = 5 − 3x on [0, 1] 50. f (x) = 9 − x 2 on [0, 3] As in Fig. 5.3.19, the region under the graph of f (x) = hx/b for 0 x b is a triangle with base b and height h. Use Eq. (7) to verify—with the notation of Eq. (16)— that n f (xi ) x = 12 bh, lim n→∞ FIGURE 5.3.16 Problem 36. 37. f (x) = x 2 on [0, 1]; n = 5 38. f (x) = x 2 on [1, 3]; n = 5 39. f (x) = 9 − x 2 on [0, 3]; n = 5 (Fig. 5.3.17) i=1 in agreement with the familiar formula for the area of a triangle. y (b, h) r y y= hx b 8 π /n O h 4 r 1 3 x 2 b x FIGURE 5.3.19 Problem 51. FIGURE 5.3.17 Problem 39. FIGURE 5.3.20 Problem 52. In Problems 52 and 53, let A denote the area and C the circumference of a circle of radius r and let A n and Cn denote the area and perimeter, respectively, of a regular n-sided polygon inscribed in this circle. 40. f (x) = 9 − x 2 on [1, 3]; n = 8 41. f (x) = x 3 on [0, 1]; n = 10 √ 42. f (x) = x on [0, 1]; n = 10 (Fig. 5.3.18) 52. Figure 5.3.20 shows one side of the n-sided polygon subtending an angle 2π/n at the center O of the circle. Show that π π π An = nr 2 sin cos and that Cn = 2nr sin . n n n y 0.8 0.4 0.2 0.4 0.6 0.8 FIGURE 5.3.18 Problem 42. 340 π /n 1 x 53. Deduce that A = 12 rC by taking the limit of An /Cn as n → ∞. Then, under the assumption that A = πr 2 , deduce that C = 2πr . Thus the familiar circumference formula for a circle follows from the familiar area formula for a circle. Riemann Sums and the Integral SECTION 5.4 341 5.4 RIEMANN SUMS AND THE INTEGRAL Suppose that f is a positive-valued and increasing function defined on a set of real numbers that includes the interval [a, b]. In Section 5.3 we used inscribed and circumscribed rectangles to set up the sums n n f (xi−1 ) x f (xi ) x and i=1 (1) i=1 that approximate the area A under the graph of y = f (x) from x = a to x = b. Recall that the notation in Eq. (1) is based on a division of the interval [a, b] into n subintervals, all with the same length x = (b − a)/n, and that [xi−1 , xi ] denotes the ith subinterval. The approximating sums in Eq. (1) are both of the form n f (xi ) x, (2) i=1 where xi denotes a selected point of the ith subinterval [xi−1 , xi ] (Fig. 5.4.1). Sums of the form in (2) appear as approximations in a wide range of applications and form the basis for the definition of the integral. Motivated by our discussion of area in Section 5.3, we want to define the integral of f from a to b as some sort of limit, as x → 0, of sums such as the one in (2). Our goal is to begin with a fairly general function f and define a computable real number I (the integral of f ) that—in the special case when f is continuous and positive-valued on [a, b]—will equal the area under the graph of y = f (x). y = f (x) a = x0 x1★ x1 x2★ x2 xi − 1 x★ i xi xn★ xn = b x -axis FIGURE 5.4.1 The Riemann sum in Eq. (2) as a sum of areas of rectangles. Riemann Sums We begin with a function f defined on [a, b] that is not necessarily either continuous or positive valued. A partition P of [a, b] is a collection of subintervals [x0 , x1 ], [x1 , x2 ], [x2 , x3 ], ... , [xn−1 , xn ] of [a, b] such that xi − 1 x★ i xi FIGURE 5.4.2 The selected point xi in the ith subinterval [xi−1 , xi ]. a = x0 < x1 < x2 < x3 < · · · < xn−1 < xn = b, as in Fig. 5.4.1. We write xi = xi − xi−1 for the length of the ith subinterval [xi−1 , xi ]. To get a sum such as the one in (2), we need a point xi in the ith subinterval for each i, 1 i n. A collection of points S = {x1 , x2 , x3 , . . . , xn } with xi in [xi−1 , xi ] for each i (Fig. 5.4.2) is called a selection for the partition P. 341 342 CHAPTER 5 The Integral DEFINITION Riemann Sum Let f be a function defined on the interval [a, b]. If P is a partition of [a, b] and S is a selection for P, then the Riemann sum for f determined by P and S is n f (xi ) xi . R= (3) i=1 We also say that this Riemann sum is associated with the partition P. The German mathematician G. F. B. Riemann (1826–1866) provided a rigorous definition of the integral. Various special types of “Riemann sums” had appeared in area and volume computations since the time of Archimedes, but it was Riemann who framed the preceding definition in its full generality. The point xi in Eq. (3) is simply a selected point of the ith subinterval [xi−1 , xi ]. That is, it can be any point of this subinterval. But when we compute Riemann sums, we usually choose the points of the selection S in some systematic manner, as illustrated in Fig. 5.4.3. There we show different Riemann sums for the function f (x) = 2x 3 − 6x 2 + 5 on the interval [0, 3]. Figure 5.4.3(a) shows rectangles associated with the left-endpoint sum n R left = f (xi−1 ) x, (4) i=1 in which each xi is selected to be xi−1 , the left endpoint of the ith subinterval [xi−1 , xi ] of length x = (b − a)/n. Figure 5.4.3(b) shows rectangles associated with the y 4 2 2 3 x 1 −2 (a) y 4 2 2 3 x 1 −2 (b) y 4 2 2 3 x 1 −2 (c) FIGURE 5.4.3 Riemann sums for f (x) = 2x 3 − 6x 2 + 5 on [0, 3]: (a) Left-endpoint sum; (b) Right-endpoint sum; (c) Midpoint sum. 342 Riemann Sums and the Integral SECTION 5.4 343 right-endpoint sum n R right = f (xi ) x, (5) i=1 in which each xi is selected to be xi , the right endpoint of [xi−1 , xi ]. In each figure, some of the rectangles are inscribed and others are circumscribed. Figure 5.4.3(c) shows rectangles associated with the midpoint sum n R mid = f (m i ) x, (6) i=1 in which xi−1 + xi , 2 the midpoint of the ith subinterval [xi−1 , xi ]. The dashed lines in Fig. 5.4.3(c) represent the ordinates of f at these midpoints. xi = m i = EXAMPLE 1 In Example 1 of Section 5.3 we calculated left- and right-endpoint sums for f (x) = x 2 on [0, 3] with n = 10 subintervals. We now do this more concisely by using summation notation, and we also calculate the analogous midpoint sum. Figure 5.4.4 shows a typical approximating rectangle for each of these sums. 3 , we see that the ith subdivision point is With a = 0, b = 3, and x = (b − a)/n = 10 xi = a + i · x = 3 i. 10 y y y y = x2 y = x2 x x i − 1 xi 3 y = x2 x x i − 1 xi 3 (a) x i − 1 xi 3 (b) x (c) FIGURE 5.4.4 Example 1: (a) The case xi = xi−1 ; (b) The case xi = xi ; (c) The case xi = m i . mi = xi − 1 = 3 (i 10 3 (2i 20 − 1) − 1) xi = 3 i 10 FIGURE 5.4.5 The ith subinterval of Example 1. The ith subinterval, as well as its midpoint xi−1 + xi 1 3i − 3 3i 3 mi = = + = (2i − 1), 2 2 10 10 20 are shown in Fig. 5.4.5. With xi = xi−1 = in Eq. (4), n R left = 3 (i 10 10 f (xi−1 ) x = i=1 3 10 − 1), we obtain the left-endpoint sum (i − 1) 2 3 10 i=1 = 27 1000 · 02 + 12 + 22 + · · · + 92 = 7695 1000 = 7.695 [using Eq. (8) of Section 5.3]. 343 344 CHAPTER 5 The Integral With xi = xi = y 3 i, 10 we get the right-endpoint sum in Eq. (5), n 0.4 π /3 1 2π /3 2 x 3 π f (xi ) x = i=1 i=1 = 27 1000 · (1 + 2 + 32 + · · · + 102 ) = 10395 1000 Finally, with xi = m i = (a) 2 2 = 10.395 3 (2i 20 = [using Eq. (8) of Section 5.3]. − 1), we get the midpoint sum in Eq. (6), n 10 R mid = y 3 2 3 i 10 10 10 R right = 0.8 f (m i ) x = i=1 27 4000 3 20 2 3 (2i − 1) 10 i=1 · 12 + 32 + 52 + · · · + 172 + 192 = 35910 4000 = 8.9775. The midpoint sum is much closer than either endpoint sum to the actual value 9 (of the area under the graph of y = x 2 over [0, 3]) that we found in Example 7 of Section 5.3. ◗ 0.8 0.4 π /3 1 2π /3 2 x 3 π (b) y 0.8 0.4 π /3 1 2π /3 2 x 3 π (c) FIGURE 5.4.6 Approximating the area under y = sin x on [0, π ] (Example 2): (a) Left-endpoint sum; (b) Right-endpoint sum; (c) Midpoint sum. EXAMPLE 2 Figure 5.4.6 illustrates Riemann sums for f (x) = sin x on [0, π] based on n = 3 subintervals: [0, π/3], [π/3, 2π/3], and [2π/3, π], of length x = π/3, and with midpoints π/6, π/2, and 5π/6. The left-endpoint sum is n π 2π π f (xi−1 ) = · sin 0 + sin + sin R left = (x) · 3 3 3 i=1 √ √ √ 3 3 π 3 π + = ≈ 1.81. = · 0+ 3 2 2 3 It is clear from the figure that the right-endpoint sum has the same value. The corresponding midpoint sum is π π 5π π 1 1 2π π = · +1+ = ≈ 2.09. Rmid = · sin + sin + sin 3 6 2 6 3 2 2 3 (We will soon be able to show that the area under one arch of the sine curve is exactly 2.) ◗ The Integral as a Limit In the case of a function f that has both positive and negative values on [a, b], it is necessary to consider the signs indicated in Fig. 5.4.7 when we interpret geometrically the Riemann sum in Eq. (3). On each subinterval [xi−1 , xi ], we have a rectangle with width xi and “height” f (xi ). If f (xi ) > 0, then this rectangle stands above the x-axis; if f (xi ) < 0, it lies below the x-axis. The Riemann sum R is then the sum of y y = f(x) x★ i + a = x0 + x1 + x2 + + + − xn − 1 xn = b xi − 1 xi − − − − − − − ★ (x★ i , f(xi )) FIGURE 5.4.7 A geometric representation of the Riemann sum in Eq. (3). 344 x Riemann Sums and the Integral SECTION 5.4 345 the signed areas of these rectangles—that is, the sum of the areas of those rectangles that lie above the x-axis minus the sum of the areas of those that lie below the x-axis. If the widths xi of these rectangles are all very small, then it appears that the corresponding Riemann sum R will closely approximate the area from x = a to x = b under y = f (x) and above the x-axis, minus the area that lies above the graph and below the x-axis. This suggests that the integral of f from a to b should be defined by taking the limit of the Riemann sums as the widths xi all approach zero: n I = lim xi →0 f (xi ) xi . (7) i=1 The formal definition of the integral is obtained by saying precisely what it means for this limit to exist. The norm of the partition P is the largest of the lengths xi = xi − xi−1 of the subintervals in P and is denoted by |P|. Briefly, Eq. (7) means that if |P| is sufficiently small, then all Riemann sums associated with the partition P are close to the number I . DEFINITION The Definite Integral The definite integral of the function f from a to b is the number n f (xi ) xi , I = lim |P|→0 (8) i=1 provided that this limit exists, in which case we say that f is integrable on [a, b]. Equation (8) means that, for each number > 0, there exists a number δ > 0 such that n f (x ) x I − i < i i=1 for every Riemann sum associated with any partition P of [a, b] for which |P| < δ. The customary notation for the integral of f from a to b, due to the German mathematician and philosopher G. W. Leibniz, is I = b n f (x) d x = lim |P|→0 a y = f(x) f(x) dx a x b FIGURE 5.4.8 Origin of Leibniz’s notation for the integral. f (xi ) xi . (9) i=1 Considering I to be the area under y = f (x) from a to b, Leibniz first thought of a narrow strip with height f (x) and “infinitesimally small” width d x (as in Fig. 5.4.8), so that its area would be the product f (x) d x. He regarded the integral as a sum of areas of such strips and denoted this sum by the elongated capital S (for summa) that appears as the integral sign in Eq. (9). We shall see that this integral notation is not only highly suggestive, but also is exceedingly useful in manipulations with integrals. The numbers a and b are called the lower limit and upper limit, respectively, of the integral; they are the endpoints of the interval of integration. The function f (x) that appears between the integral sign and d x is called the integrand. The symbol d x that follows the integrand in Eq. (9) should, for the time being, be thought of as simply an indication of what the independent variable is. Like the index of summation, the independent variable x is a “dummy variable”—it may be replaced with any other variable without affecting the meaning of Eq. (9). Thus if f is integrable on [a, b], we can write b b b f (x) d x = f (t) dt = f (u) du. a a a The definition given for the definite integral applies only if a < b, but it is convenient to include the cases a = b and a > b as well. The integral is defined in 345 346 CHAPTER 5 The Integral these cases as follows: a f (x) d x = 0 (10) a and b f (x) d x = − a a f (x) d x, (11) b provided that the right-hand integral exists. Thus interchanging the limits of integration reverses the sign of the integral. Just as not all functions are differentiable, not every function is integrable. Suppose that c is a point of [a, b] such that f (x) → +∞ as x → c. If [xk−1 , xk ] is the subinterval of the partition P that contains c, then the Riemann sum in Eq. (3) can be made arbitrarily large by choosing xk to be sufficiently close to c. For our purposes, however, we need to know only that every continuous function is integrable. The following theorem is proved in Appendix G. THEOREM 1 Existence of the Integral If the function f is continuous on [a, b], then f is integrable on [a, b]. Although we omit the details, it is not difficult to show that the definition of the integral can be reformulated in terms of sequences of Riemann sums, as follows. THEOREM 2 The Integral as a Limit of a Sequence The function f is integrable on [a, b] with integral I if and only if lim Rn = I (12) n→∞ for every sequence {Rn }∞ 1 of Riemann sums associated with a sequence of partitions |Pn | → 0 as n → +∞. of [a, b] such that {Pn }∞ 1 Riemann Sum Computations The reformulation in Theorem 2 of the definition of the integral is helpful because it is easier to visualize a specific sequence of Riemann sums than to visualize the vast totality of all possible Riemann sums. In the case of a continuous function f (known to be integrable by Theorem 1), the situation can be simplified even more by using only Riemann sums associated with partitions consisting of subintervals all with the same length x1 = x2 = · · · = xn = b−a = x. n Such a partition of [a, b] into equal-length subintervals is called a regular partition of [a, b]. Any Riemann sum associated with a regular partition can be written in the form n f (xi ) x, (13) i=1 where the absence of a subscript in x signifies that the sum is associated with a regular partition. In such a case the conditions |P| → 0, x → 0, and n → +∞ are 346 Riemann Sums and the Integral SECTION 5.4 347 equivalent, so the integral of a continuous function can be defined quite simply: b n f (x) d x = lim n→∞ a n f (xi ) x = lim x→0 i=1 f (xi ) x. (14) i=1 Consequently, we henceforth will use only regular partitions; the subintervals will thus have length and endpoints given by b−a n x = xi = a + i · x and (15) for i = 0, 1, 2, 3, . . . , n. If we select xi = xi then (14) gives b n f (x) d x = lim n→∞ a f (xi ) x. (16) i=1 4 EXAMPLE 3 Use Riemann sums to evaluate (x 3 − 2x) d x. 0 Solution With a = 0 and b = 4 in (15), we have x = 4/n and xi = 4i/n. Hence 4 n n 4i 4i 3 4 3 (x − 2x) d x = lim f (xi ) x = lim −2 · n→∞ n→∞ n n n 0 i=1 i=1 n 3 64i 8i 4 − = lim 3 n→∞ n n n i=1 256 n 3 32 n = lim i − 2 i . n→∞ n 4 i=1 n i=1 We now use Eqs. (7) and (9) in Section 5.3 to convert each of the last two sums to closed form: 4 32 n(n + 1) 256 n 2 (n + 1)2 (x 3 − 2x) d x = lim · · − n→∞ n4 4 n2 2 0 16 64 (n + 1)2 − (n + 1) = lim n→∞ n 2 n 2 1 1 − 16 1 + ; = lim 64 1 + n→∞ n n 4 (x 3 − 2x) d x = 64 − 16 = 48. ◗ 0 EXAMPLE 4 Use Riemann sums to evaluate b x d x (where a < b). a Solution With f (x) = x and xi = xi (see Fig. 5.4.9), Eqs. (15) and (16) yield b n n x d x = lim f (xi ) x = lim (a + i · x) x a n→∞ = lim n→∞ = lim n→∞ n→∞ i=1 (a x) i=1 n n 1 + (x)2 i=1 i i=1 n(n + 1) (a x) · n + (x) · 2 2 347 348 CHAPTER 5 The Integral y y=x f(xi ) = xi a = x0 xi − 1 xi Δx b = xn x FIGURE 5.4.9 Calculating the area under y = x from x = a to x = b. when we use Eqs. (6) and (7) in Section 5.3 to convert the sums to closed form. Then substituting x = (b − a)/n gives b b − a 2 n(n + 1) b−a x d x = lim a · · ·n+ n→∞ n n 2 a 1 1 = lim a(b − a) + (b − a)2 1 + n→∞ 2 n 1 1 1 = a(b − a) + (b − a)2 = (b − a) a + b − a 2 2 2 1 1 = (b − a) · (b + a) = (b2 − a 2 ). 2 2 y y=x + a − b x Thus we see finally that b x dx = a REMARK 1 If 0 < a < b, then A = Fig. 5.4.9. Then Eq. (17) implies that FIGURE 5.4.10 Example 4 with a < 0 < b. b a 1 2 1 2 b − a . 2 2 (17) ◗ x d x is the area of the trapezoid shown in A = (b − a) · 12 (a + b) = w · h, where w = b − a is the width and h = 12 (a + b) is the average height of the trapezoid. y y=x − a b FIGURE 5.4.11 Example 4 with 0 < a < b. x REMARK 2 Figures 5.4.10 and 5.4.11 illustrate two different cases in Example 4. In each case Eq. (17) agrees with the sum of the indicated signed areas. The minus sign in Fig. 5.4.10 represents the fact that area beneath the x-axis is measured with a negative number. The minus sign in Fig. 5.4.11 signifies that the area of the triangle over [0, a] is subtracted from the area of the triangle over [0, b] to get the area of the trapezoid. The summation formulas in Eqs. (6) through (9) in Section 5.3 suffice for the integration of polynomials of low degree, but integrals of other functions may require other devices (or a computer algebra system) for the conversion of Riemann sums to closed forms whose limits can be evaluated. 2 EXAMPLE 5 Use Riemann sums to evaluate e x d x. 0 348 Riemann Sums and the Integral SECTION 5.4 349 Solution With x = 2/n and xi = i · x = 2i/n, we have 2 n n 2 e x d x = lim f (xi ) x = lim e2i/n · n→∞ n→∞ n 0 i=1 i=1 2 2/n e + e4/n + e6/n + · · · + e2n/n = lim n→∞ n 2e2/n = lim 1 + e2/n + e4/n + · · · + e2(n−1)/n n→∞ n 2r 1 + r + r 2 + · · · + r n−1 = lim n→∞ n 2/n where r = e . To convert the last sum to closed form, we use the formula rn − 1 , (18) 1 + r + r 2 + · · · + r n−1 = r −1 which is readily verified by multiplying the left-hand side by the denominator on the right. This gives 2 2r r n − 1 · e x d x = lim n→∞ n r −1 0 2/n e2 − 1 2(e2 − 1) 2e · 2/n = . (19) = lim n→∞ n e −1 lim n · (1 − e−2/n ) n→∞ The limit in the denominator has the indeterminate form ∞ · 0 as n → +∞. We evaluate it using l’Hôpital’s rule as in Section 4.9: 1 − e−2/n [now the indeterminate form 0/0] 1 n→∞ n→∞ n 2 − 2 · e−2/n n = lim = lim 2e−2/n = 2. 1 n→∞ n→∞ − 2 n Substituting this limit in Eq. (19) finally gives 2 e x d x = e2 − 1. ◗ lim n · (1 − e−2/n ) = lim 0 5.4 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. A partition P of [a, b] is a collection of subintervals [x0 , x1 ], [x1 , x2 ], [x2 , x3 ], ..., [xn−1 , xn ] of [a, b] such that a = x0 < x1 < x2 < x3 < · · · < xn−1 < xn = b. 2. If P is a partition of [a, b], then a selection S for P is a collection of points S = {x1 , x2 , x3 , . . . , xn } such that xi−1 xi xi for 1 i n. 3. If f is a function defined on [a, b], P is a partition of [a, b], and S is a selection for P (with the same notation as in Questions 1 and 2), then the Riemann sum for f determined by P and S is n R= f (xi )(xi − xi−1 ). i=1 349 350 CHAPTER 5 The Integral 4. The midpoint sum with n = 10 subintervals for f (x) = x 2 on [0, 3] is 8.9775. 2π 5. The midpoint sum for f (x) = sin x with n = 3 subintervals of [0, π] is . 3 6. The norm |P| of the partition P = {x0 , x1 , x2 , . . . , xn } is the maximum value of xi = xi − xi−1 for 1 i n. 7. Suppose that f is a function defined on [a, b]. In the notation of Questions 1, 2, and 6, the definite integral of f from a to b is n I = lim |P|→0 f (xi )xi i=1 provided that this limit exists. 8. A common notation for the definite integral of f from a to b is b 9. If f is continuous on [a, b], then f (x) d x exists. a b 1 1 10. Example 4 shows that x d x = b2 − a 2 . 2 2 a b f (x) d x. a 5.4 CONCEPTS: QUESTIONS AND DISCUSSION 1. Explain why you would generally expect the midpoint sum in Eq. (6) to be a b more accurate approximation to the actual value a f (x) d x than either the leftendpoint sum in (4) or the right-endpoint sum in (5). 2. The result in Example 4, with a = 0, and Problems 49 and 50 tell us that b b b 1 1 1 x d x = b2 , x 2 d x = b3 , and x 3 d x = b4 2 3 4 0 0 0 if b > 0. Assuming that the pattern holds (it does), what would you expect to be b the value of a x n d x with n > 0 and 0 < a < b? Explain how you take into account the nonzero lower limit a. 2 5 3. Example 5 and Problem 56 imply that 0 e x d x = e2 − 1 and 0 e x d x = e5 − 1. Thinking of area under the curve y = e x , what would you expect to be the b 5 value of 2 e x d x? What would you conjecture about the value of a e x d x with 0 < a < b? 5.4 PROBLEMS n In Problems 1 through 10, express the given limit as a definite integral over the indicated interval [a, b]. Assume that [xi−1 , xi ] denotes the ith subinterval of a subdivision of [a, b] into n subintervals, all with the same length x = (b − a)/n, and that m i = 12 (xi−1 + xi ) is the midpoint of the ith subinterval. 5. lim n→∞ n n (2xi − 1)x 1. lim n→∞ [1, 3] over i=1 n n (2 − 3xi−1 )x 2. lim n→∞ n 4. lim 350 xi2 + 4 x i=1 xi3 − 3xi2 i=1 over [0, 5] 1 x √ 1 + mi over [3, 8] (cos 2xi−1 ) x over (sin 2π m i ) x over [0, 1/2] e2xi x [0, 1] n 8. lim n→∞ [0, π/2] i=1 n over [0, 10] 9. lim n→∞ i=1 n n→∞ [−3, 2] i=1 3. lim n→∞ over 25 − xi2 x i=1 7. lim n→∞ over [4, 9] i=1 6. lim n→∞ √ m i x + 1 x i=1 n over [0, 3] 10. lim n→∞ i=1 over Riemann Sums and the Integral SECTION 5.4 In Problems 11 through 20, compute the Riemann sum n f (xi ) x i=1 for the indicated function and a regular partition of the given interval into n subintervals. Use xi = xi , the right-hand endpoint of the ith subinterval [xi−1 , xi ]. 11. f (x) = x 2 on [0, 1]; n = 5 12. f (x) = x 3 on [0, 1]; n = 5 1 13. f (x) = on [1, 6]; n = 5 x √ 14. f (x) = x on [0, 5]; n = 5 15. f (x) = 2x + 1 on [1, 4]; n = 6 16. f (x) = x 2 + 2x on [1, 4]; n = 6 17. f (x) = x 3 − 3x on [1, 4]; n = 5 √ 18. f (x) = 1 + 2 x on [2, 3]; n = 5 19. f (x) = cos x on [0, π]; n = 6 20. f (x) = ln x on [1, 6]; n = 5 21. through 30. Repeat Problems 11 through 20, except with xi = xi−1 , the left-hand endpoint. 31. through 40. Repeat Problems 11 through 20, except with xi = (xi−1 + xi )/2, the midpoint of the ith subinterval. 41. Work Problem 13 with xi = (3xi−1 + 2xi )/5. 42. Work Problem 14 with xi = (xi−1 + 2xi )/3. In Problems 43 through 48, evaluate the given integral by computing n f (xi ) x lim n→∞ i=1 for a regular partition of the given interval of integration. 4 2 x2 dx 44. x3 dx 43. 0 0 3 (2x + 1) d x 45. a a [Suggestion: First consider the case a < b.] 54. Suppose that the function f is defined on the interval [0, 1] as follows: ⎧ ⎨ 1 if 0 < x 1, f (x) = x ⎩ 0 if x = 0. 1 Show that the integral 0 f (x) d x does not exist. [Suggestion: Show that, whatever n may be, the first term in the n Riemann sum i=1 f (xi ) x can be made arbitrarily large by the choice of the first selected point xi .] Why does this not contradict Theorem 1? 55. Suppose that the function f is defined as follows: 0 if x is rational, f (x) = 1 if x is irrational. 1 Show that the integral 0 f (x) d x does not exist. [Suggestion: Show that, whatever n may be, the Riemann sum n i=1 f (x i ) x has the value 0 for one possible selection of points {xi }, but the value 1 for another possible selection.] Why does this not contradict Theorem 1? Use the method of Example 5 to verify the results in Problems 56 through 58. 5 3 56. e x d x = e5 − 1. 57. e−x d x = 1 − e−3 . 5 (3x + 1) d x 2 48. 0 0 0 5 e dx = e − e . x 5 (4 − 3x) d x 58. (x 3 − x) d x 59. First show that π 4 49. Show by the method of Example 4 that b 1 x 2 d x = b3 3 0 i=1 xi xi = 1 2 1 2 b − a . 2 2 Explain why this computation proves that b b2 − a 2 . x dx = 2 a sin x d x = lim n→∞ 0 π n n sin k=1 kπ . n A computer algebra system reports that n sin if b > 0. 50. Show by the method of Example 4 that b 1 x 3 d x = b4 4 0 n 2 2 0 if b > 0. 51. Let f (x) = x, and let {x0 , x1 , x2 , . . . , xn } be an arbitrary partition of the closed interval [a, b]. For each i (1 i n), let xi = (xi−1 + xi )/2. Then show that a 53. Suppose that f (x) ≡ c, a constant. Use Riemann sums to prove that b c d x = c(b − a). 1 3 47. 52. Suppose that f is a function continuous on [a, b] and that k is a constant. Use Riemann sums to prove that b b k f (x) d x = k f (x) d x. 0 46. 351 k=1 π kπ = cot . n 2n Use this fact and l’Hôpital’s rule to show finally that π sin x d x = 2. 0 In Problems 60 through 62, verify the given result as follows: Use a computer algebra system first to set up the appropriate Riemann sum, then to simplify the sum, and finally to evaluate its limit as n → +∞. b 60. e x d x = eb − ea . a b sin x d x = cos a − cos b. 61. a b cos x d x = sin b − sin a. 62. a 351 352 CHAPTER 5 The Integral 5.4 INVESTIGATION: Calculator/Computer Riemann Sums Suppose that you want to approximate the integral b f (x) d x a numerically using midpoint sums. If x = (b − a)/n and 1 1 1 m i = xi − x = (a + i · x) − x = a + i − x 2 2 2 is the midpoint of the ith subinterval [xi−1 , xi ], then the selection xi = m i in Eq. (14) gives b n f (x) d x = lim n→∞ a f (m i ) x. i=1 Many calculators and computer algebra systems include Sum commands that can be used to calculate easily and rapidly the midpoint sum with larger and larger values of n. A common practice is to begin with perhaps n = 50 subintervals, then calculate midpoint sums with successively doubled numbers of subintervals; that is, with n = 50, 100, 200, . . . , until successive sums agree to the desired number of decimal places of accuracy. In the Project Manual material for this investigation we illustrate this procedure using graphing calculators and typical computer algebra systems. You can then carry out the following investigations. 1. Approximate the integral 2 e x d x = e2 − 1 ≈ 6.3891 0 of Example 5 accurate to four decimal places. 2. Approximate the integral π sin x d x = 2 = 2.0000 y 0 of Problem 59 accurate to four decimal places. 3. First explain why Fig. 5.4.12 and the area formula A = πr 2 for a circle of radius r imply that y = 1 − x2 1 ! 4 1 − x 2 d x = π. 0 1 FIGURE 5.4.12 Investigation 3. x Then use midpoint sums to approximate this integral and, thereby, the numerical value of π . Begin with n = 50 subintervals, then successively double n. How large must n be for you to obtain the familiar four-place approximation π ≈ 3.1416? 5.5 EVALUATION OF INTEGRALS The evaluation of integrals by using Riemann sums, as in Section 5.4, is tedious and time-consuming. Fortunately, we will seldom find it necessary to evaluate an integral in this way. In 1666, Isaac Newton, while still a student at Cambridge University, discovered a much more efficient way to evaluate an integral. A few years later, Gottfried Wilhelm Leibniz, working with a different approach, discovered this method independently. 352 Evaluation of Integrals SECTION 5.5 353 Newton’s key idea was that to evaluate the number b f (x) d x, y y = f(x) a we should first introduce the function A(x) defined as follows: A(x) = x f (t) dt. (1) a A(x) a Δ A f(x) x x + Δx b x-axis FIGURE 5.5.1 The area function A(x). The independent variable x appears as the upper limit of the integral in Eq. (1); the dummy variable t is used in the integrand merely to avoid confusion. If f is positivevalued, continuous, and x > a, then A(x) is the area below the curve y = f (x) above the interval [a, x] (Fig. 5.5.1). It is apparent from Fig. 5.5.1 that A(x) increases as x increases. When x increases by x, A increases by the area A of the narrow strip in Fig. 5.5.1 with base [x, x + x]. If x is very small, then the area of this strip is very close to the area f (x) x of the rectangle with base [x, x + x] and height f (x). Thus A ≈ f (x). (2) x Moreover, the figure makes it plausible that we get equality in the limit at x → 0: A ≈ f (x) x; dA A = lim = f (x). x→0 dx x That is, A (x) = f (x), (3) so the derivative of the area function A(x) is the curve’s height function f (x). In other words, Eq. (3) implies that A(x) is an antiderivative of f (x). Figure 5.5.2 shows a physical interpretation of Eq. (3). A paint roller is laying down a 1-mm-thick coat of paint to cover the region under the curve y = f (t). The paint roller is of adjustable length—as it rolls with a speed of 1 mm/s from left to right, one end traces the x-axis and the other end traces the curve y = f (t). At any time t, the volume V of paint the roller has laid down equals the area of the region already painted: V = A(t) (mm3 ). Then Eq. (3) yields dV = A (t) = f (t). dt Thus the instantaneous rate at which the roller is depositing paint is equal to the current length of the roller. y y= f (t) a x t FIGURE 5.5.2 The adjustable-length paint roller. 353 354 CHAPTER 5 The Integral The Evaluation Theorem Equation (3) implies that the area function A(x) defined in (1) and illustrated in Fig. 5.5.1 is one antiderivative of the given function f (x). Now suppose that G(x) is any other antiderivative of f (x)—perhaps one found by the methods of Section 5.2. Then A(x) = G(x) + C, (4) because (by the second corollary to the mean value theorem) two antiderivatives of the same function (on an interval) can differ only by a constant. Also, a f (t) dt = 0 (5) A(a) = a and A(b) = b f (t) dt = a b f (x) d x (6) a by Eq. (1). So it follows that b f (x) d x = A(b) − A(a) = [G(b) + C] − [G(a) + C], a and thus b f (x) d x = G(b) − G(a). (7) a Our intuitive discussion has led us to the statement of Theorem 1. THEOREM 1 Evaluation of Integrals If G is an antiderivative of the continuous function f on the interval [a, b], then b f (x) d x = G(b) − G(a). (7) a In Section 5.6 we will fill in the details of the preceding discussion, thus giving a rigorous proof of Theorem 1 (which is part of the fundamental theorem of calculus). Here we concentrate on the computational applications of this theorem. The difference G(b) − G(a) is customarily abbreviated as [G(x)]ab , so Theorem 1 implies that b b f (x) d x = G(x) = G(b) − G(a) a (8) a if G is any antiderivative of the continuous function f on the interval [a, b]. Thus if we can find an antiderivative G of f , we can quickly evaluate the integral without having to resort to the paraphernalia of limits of Riemann sums. If G (x) = f (x), then (as in Section 5.2) we write f (x) d x = G(x) + C (9) for the indefinite integral of f . With the indefinite integral antiderivative G(x), Eq. (8) takes the form b b f (x) d x = f (x) d x . a f (x) d x in place of the (10) a This is the connection between the indefinite integral and the definite integral to which we have alluded in the earlier sections of Chapter 5. 354 Evaluation of Integrals SECTION 5.5 EXAMPLE 1 Because xn dx = x n+1 +C n+1 355 (if n = −1), it follows that b x n+1 x dx = n+1 b = n a if n = −1. For instance, 3 x dx = 2 3 = 1 3 x 3 0 a bn+1 − a n+1 n+1 · 33 − 1 3 0 1 3 · 03 = 9. Contrast the immediacy of this result with the complexity of the computations of Ex◗ ample 7 in Section 5.3. EXAMPLE 2 Because cos x d x = sin x + C, it follows that b b cos x d x = sin x a = sin b − sin a. a Similarly, b sin x d x = b − cos x = cos a − cos b. a a In particular, as we mentioned in Example 2 of Section 5.4, π π sin x d x = − cos x = (−cos π ) − (−cos 0) = (+1) − (−1) = 2. 0 ◗ 0 EXAMPLE 3 2 x dx = 5 0 9 1 2 = 1 6 x 6 0 64 6 −0= 32 . 3 9 2x − x −1/2 − 3 d x = x 2 − 2x 1/2 − 3x = 52. 1 (2x + 1)3 d x = 0 0 1 (2x 8 + 1)4 π/2 sin 2x d x = 1 − 12 cos 2x e2x d x = 0 1 1 1 1 2x e 2 0 = 0 π/2 0 1 8 · (81 − 1) = 10. = − 12 (cos π − cos 0) = 1. = 12 (e2 − 1). We have not shown the details of finding the antiderivatives, but you can (and should) check each of these results by showing that the derivative of the function within the evaluation brackets on the right is equal to the integrand on the left. In Example 4 ◗ we show the details. 5 √ EXAMPLE 4 Evaluate 3x + 1 d x. 1 355 356 CHAPTER 5 The Integral Solution We apply the antiderivative form of the generalized power rule, u k+1 u k du = + C (k = −1), k+1 with k = 1 2 and u = 3x + 1, This gives du = 3 d x. (3x + 1)1/2 d x = 1 3 = 1 3 (3x + 1)1/2 (3 d x) = · u 3/2 3 2 1 3 u 1/2 du + C = 29 (3x + 1)3/2 + C for the indefinite integral, so it follows from Eq. (10) that 5 5 √ 3/2 2 3x + 1 d x = 9 (3x + 1) 1 1 = 2 (163/2 9 −4 3/2 ) = 29 (43 − 23 ) = 112 . 9 ◗ If the derivative F (x) of the function F(x) is continuous, then the evaluation theorem, with F (x) in place of f (x) and F(x) in place of G(x), yields b b F (x) d x = F(x) = F(b) − F(a). a (11) a The next example provides an immediate application. EXAMPLE 5 Suppose that an animal population P(t) initially numbers P(0) = 100 and that its rate of growth after t months is given by P (t) = 10 + t + (0.06)t 2 . What is the population after 10 months? Solution By Eq. (11), we know that 10 10 P(10) − P(0) = P (t) dt = [10 + t + (0.06)t 2 ] dt 0 0 10 3 1 2 = 170. = 10t + 2 t + (0.02)t 0 Thus P(10) = 100 + 170 = 270 individuals. ◗ EXAMPLE 6 Evaluate n lim n→∞ i=1 2i n2 by recognizing this limit as the value of an integral. Solution If we write n i=1 2i = n2 n i=1 2i n 1 , n we recognize that we have a Riemann sum for the function f (x) = 2x associated with a partition of the interval [0, 1] into n equal-length subintervals. The ith point of 356 Evaluation of Integrals SECTION 5.5 357 subdivision is xi = i/n, and x = 1/n. Hence it follows from the definition of the integral and from the evaluation theorem that n lim n→∞ i=1 n n 2i = lim 2x x = lim f (xi ) x i n→∞ n 2 n→∞ i=1 i=1 1 1 f (x) d x = 2x d x. = 0 0 Therefore, n lim n→∞ i=1 1 2i = x 2 = 1. n2 0 ◗ Basic Properties of Integrals Problems 59 through 62 outline elementary proofs of the integral properties that are stated next. We assume throughout that each function mentioned is integrable on [a, b]. y Integral of a Constant y=c c b ∫a c dx a b c d x = c(b − a). a b FIGURE 5.5.3 The integral of a constant is the area of a rectangle. x This property is intuitively obvious because the area represented by the integral is simply a rectangle with base b − a and height c (Fig. 5.5.3). Constant Multiple Property b c f (x) d x = c a b f (x) d x. a Thus a constant can be “moved across” the integral sign. For example, π/2 0 π/2 2 sin x d x = 2 π/2 sin x d x = 2 − cos x = 2. 0 Sum Property a b 0 [ f (x) + g(x)] d x = a b f (x) d x + b g(x) d x. a Thus if the functions f and g are both integrable on [a, b], then the integral of their sum is equal to the sum of their integrals. This fact sometimes permits a “divideand-conquer” strategy for the calculation of integrals: π π π √ √ x x 3 x dx + cos d x dx = 3 x + cos 2 2 0 0 0 π π x = 2π 3/2 + 2. = 2x 3/2 + 2 sin 2 0 0 Figure 5.5.4 illustrates geometrically the sum property of integrals. The proof of the sum property illustrates a Riemann sums approach that can be adapted to all of the properties under discussion here. Let us think of a partition of the interval [a, b] into subintervals all having the same length x. If the functions f , g, and f + g are all 357 358 CHAPTER 5 The Integral y y = f(x) + g(x) b g(x) ∫a g(x) dx f(x) ∫a a b y = f (x) f(x) dx x x b FIGURE 5.5.4 The integral of the sum of two positive-valued functions. integrable, then Theorem 2 in Section 5.4 gives b n [ f (x) + g(x)] d x = lim x→0 a f (xi ) + g(xi ) x i=1 n f (xi ) x = lim x→0 i=1 = = y = f (x) ∫a c b i=1 f (x) d x + a ∫c f (x) dx a b f (x) dx c b x FIGURE 5.5.5 The way the interval union property works. i=1 f (xi ) x + lim x→0 g(xi ) x + n n n lim x→0 g(xi ) x i=1 b g(x) d x. a Interval Union Property If a < c < b, then b f (x) d x = a c b f (x) d x + a f (x) d x. c Figure 5.5.5 indicates the plausibility of the interval union property. EXAMPLE 7 If f (x) = 2 |x|, then −2x f (x) = 2x The graph of f is shown in Fig. 5.5.6. An antiderivative of f (x) is not evident, but the interval union property allows us to split the integral of f on [−1, 3] into two easily calculated integrals: y −2x 2x y= y= 3 −1 −1 3 FIGURE 5.5.6 The area under the graph of y = 2|x| over [−1, 3]. x 2 |x| d x = 0 −1 (−2x) d x + 0 = −x 2 −1 3 (2x) d x 0 3 + x2 = [0 − (−1)] + [9 − 0] = 10. 0 ◗ Does the result agree with Fig. 5.5.6? EXAMPLE 8 Evaluate the integral 358 if x 0, if x 0. 2π 0 |cos x − sin x| d x. Evaluation of Integrals SECTION 5.5 Solution Figure 5.5.7 shows the graph of the function f (x) = cos x − sin x and Fig. 5.5.8 shows the graph of its absolute value | f (x)| = |cos x − sin x| that we want to integrate. We readily see that f (x) = 0 at x = π/4 and x = 5π/4, so ⎧ ⎨cos x − sin x if 0 x < π/4, | f (x)| = sin x − cos x if π/4 x < 5π/4, ⎩cos x − sin x if 5π/4 x 2π. 2 1.5 1 y = cos x − sin x 0.5 0 y π 4 −0.5 5π 4 −1 The interval union property therefore gives 2π |cos x − sin x| d x 0 π/4 5π/4 = (cos x − sin x) d x + (sin x − cos x) d x + −1.5 −2 0 1 2 3 x 4 5 6 FIGURE 5.5.7 y = cos x − sin x. 0 2 y = |cos x − sin x| 1.5 = sin x + cos x π 4 −0.5 5π/4 + − cos x − sin x π/4 2π (cos x − sin x) d x 5π/4 2π + sin x + cos x 5π/4 √ √ √ = 2 2+ 2 − (0 + 1) + 12 2 + 12 2 √ √ √ √ − − 12 2 − 12 2 + (0 + 1) − − 12 2 − 12 2 √ = 4 2. 0.5 0 π/4 π/4 0 1√ 1 y 359 5π 4 −1 1 2 ◗ −1.5 −2 0 1 2 3 x 4 5 6 FIGURE 5.5.8 y = | cos x − sin x|. Comparison Properties (1) If f (x) g(x) for all x in [a, b], then b f (x) d x a b g(x) d x. a (2) If m f (x) M for all x in [a, b], then b f (x) d x M(b − a). m(b − a) y a M y = f (x) m a b−a x b EXAMPLE 9 Figure 5.5.10 shows the graphs √ √ y = 1+ x , y = 1+x, FIGURE 5.5.9 Plausibility of the second comparison property. and we see that 2 y = 1.2 + 0.3x y= 1+ x y 1 0 The first comparison property says that the larger function has the larger integral. The plausibility of the second comparison property is indicated in Fig. 5.5.9. Note that m and M need not necessarily be the minimum and maximum values of f (x) on [a, b ]. y= 1+x 0.5 x 1 y = 1.2 + (0.3)x, and √ √ 1 + x 1 + x 1.2 + (0.3)x (12) √ √ 1+x for [0, 1]. Indeed, the fact that x x for x in [0, 1] implies that ! x in ! √ √ 1 + x there. The graph y = 1.2 + (0.3)x that lies above y = 1 + x was discovered empirically using a graphing calculator. At any rate, the inequalities in (12) and the first comparison property of integrals imply that 1 1 1 √ √ 1 + x dx 1 + x dx [1.2 + (0.3)x] d x; 0 0 0 thus FIGURE 5.5.10 ! Bounding the √ graph of f (x) = 1 + x. 1 2 (1 3 + x) 3/2 0 0 1 1 √ 2 1 + x d x (1.2)x + (0.15)x = 1.35. 0 359 360 CHAPTER 5 The Integral Now 23 (23/2 − 1) ≈ 1.2190, so we see finally that 1.21 1 √ 1 + x d x 1.35. (13) 0 1! √ It turns out (using the methods of Section 5.9) that the actual value of 0 1 + x d x is 1.29 rounded to two decimal places—quite close to the average 1.28 of the upper ◗ and lower bounds in (13). The properties of integrals stated here are frequently used in computing and will be applied in the proof of the fundamental theorem of calculus in Section 5.6. 5.5 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. x f (t) dt, then A (x) = f (t). 1. If A(x) = a b 2. If G (x) = f (x) for x in [a, b], then f (x) d x = G(b) − G(a). a 3 1 x 2 d x = (27 − 0) = 9. 3. 3 0 b 4. cos x d x = sin a − sin b. a 2 32 . x5 dx = 5. 3 0 b 6. If F is continuous on [a, b], then F (x) d x = F(b) − F(a). a 7. If f and g are integrable on [a, b], then b [ f (x) + g(x)] d x = a b f (x) d x + b g(x) d x. a a 8. If f is integrable on [a, b] and a < c < b, then b f (x) d x = a c b f (x) d x + a f (x) d x. c √ | cos x − sin x| d x = 4 2. 0 1 √ 1 + x d x 1.35. 10. 1.21 2π 9. 0 5.5 CONCEPTS: QUESTIONS AND DISCUSSION 1. Let f be a continuous function defined on the closed interval [a, b]. Explain the difference between the functions g and h defined for t in [a, b] by g(t) = a b f (x) d x and h(t) = t f (x) d x. a What is the difference between their derivatives g (t) and h (t)? 360 Evaluation of Integrals SECTION 5.5 2. What is the relation between the integrals b f (x) d x and a b 361 | f (x)| d x? a What is the difference between their absolute values? Discuss separately the following cases: (a) f is positive-valued on the interval [a, b]; (b) f is negative-valued on the interval [a, b]; (c) f has both positive and negative values on [a, b]. 5.5 PROBLEMS Apply the evaluation theorem to evaluate the integrals in Problems 1 through 36. 1 2 √ √ 3x + 2 x + 3 3 x d x 1. 0 3 2. 1 −1 −2 (x − x ) d x 6. 8. 1 4 10. 1 x +1 dx x2 x 99 d x 13. −1 4 4 5/2 7x − 5x 3/2 d x (3x 2 + 2x + 4) d x x 99 d x 0 1 3 3 18. 1 9 20. 10 dx (2x + 3)2 √ 2 1+ x dx 22. √ −1 8 x 2/3 d x 19. 1 1 21. −1 3t dt (2x + 1) d x 3 1 4 0 17. 1 2 23. 0 (e x − e−x ) d x √ e3t dt 3 0 0 π/4 29. sin x cos x d x 0 0 π 31. sin 5x d x sin2 x cos x d x 34. 5 sin 0 5 πx dx 10 (2 − |x|) d x −3 6 46. 0 5 − |2x| d x 0 5 47. ! 25 − x 2 d x 0 6 48. ! 6x − x 2 d x (Suggestion: Complete the square.) 0 In Problems 49 through 54, use properties of integrals to establish each inequality without evaluating the integrals involved. 1! 1 √ 2 49. 1 1 + x dx 1 + x dx cos πt dt i=1 πi 1 sin n n −2 45. 2 i2 n3 b In Problems 43 through 48, an integral a f (x) d x is given. First sketch the graph y = f (x) on the interval [a, b ]. Then, interpreting the integral as the area of a region, evaluate it using known area formulas for rectangles, triangles, and circles. 2 3 |1 − x| d x |3x − 2| d x 43. 44. 0 2 50. √ 1 + x dx 1 0 cos 3x d x 0 n→∞ 0 32. π/2 33. π 30. i=1 42. lim 0 1 du Note the abbreviation for 24. du. u2 u2 2 2 10 1 1 25. dt 26. dx t x 1 5 π/2 1 (e x − 1)2 d x 28. cos 2x d x 27. n→∞ n 3 (x − 1)5 d x 2 In Problems 37 through 42, evaluate the given limit by first recognizing the indicated sum as a Riemann sum associated with a regular partition of [0, 1] and then evaluating the corresponding integral. n 2i 1 37. lim −1 n→∞ n n i=1 √ x dx sec2 2t dt 0 1 + 2 + 3 + ··· + n 39. lim n→∞ n2 3 3 1 + 2 + 33 + · · · + n 3 40. lim n→∞ n4 √ √ √ √ 1 + 2 + 3 + ··· + n 41. lim √ n→∞ n n 1 15. π/8 36. (x + 1) d x 2 −1 0 πx dx 4 38. lim 3 cos n 0 0 11. (x + 1) d x (x 4 − x 3 ) d x −1 16. 1 9. 1 √ dx x 2 0 4 0 14. 7. 1 12. 3 1 3 x 3 (1 + x)2 d x 0 5. 2 4 1 3. 1 dx x4 4. 6 dx x2 2 35. 51. 0 0 2 ! √ 1 + x 3 d x 10 1 1 1 √ dx 1+ x 1 0 1 dx 1 + x2 361 362 CHAPTER 5 The Integral 5 1 1 d x dx 5 2 2 1+x 2 1+x 2 √ sin x d x 2 53. 5 (whose exact value is known to be ln 2 ≈ 0.693). 52. 0 54. π 8 0 π/4 1.2 1 π dx 2 1 + cos x 6 1 x 0.8 y In Problems 55 through 58, use the second comparison property of integrals to estimate—giving both a lower bound and an upper bound as in Problem 54—the value of the given integral. 9 1 1 1 dx 56. 55. √ dx 1 + x 1 + x 0 4 π/4 ! π/6 cos2 x d x 58. 16 + 2 sin2 x d x 57. 0 y= 1 0.4 0.2 0 62. Use sequences of Riemann sums to establish the interval union property of the integral. Note that if Rn and Rn are Riemann sums for f on the intervals [a, c] and [c, b], respectively, then Rn = Rn + Rn is a Riemann sum for f on [a, b]. 63. Suppose that a tank initially contains 1000 gal of water and that the rate of change of its volume after the tank drains for t min is V (t) = (0.8)t − 40 (in gallons per minute). How much water does the tank contain after it has been draining for a half-hour? 1.2 1.4 1.6 1.8 2 FIGURE 5.5.11 Bounding the 1 graph of f (x) = . x 59. Use Riemann sums—as in the proof of the sum property of integrals—to establish the constant multiple property. 61. Deduce the second comparison property of integrals from the first comparison property. 1 x 0 60. Use Riemann sums to establish the first comparison property of integrals. ( 32 , 23 ) 0.6 66. Figure 5.5.12 shows the graph of f (x) = 1/(1 + x 2 ) on the interval [0, 1], the line y = L(x) joining its endpoints (0, 1) and (1, 12 ), and the line y = L(x) + 0.07. First graph f (x) − L(x) to verify that the latter line lies above y = f (x) on the interval [0, 1]. Then use this construction to estimate the value of the integral 1 1 dx 2 0 1+x (whose exact value is known to be 14 π ≈ 0.785). 1.2 y = L(x) + 0.07 y= 1 1 1 + x2 0.8 y = L(x) 64. Suppose that the population of Juneau in 1970 was 125 (in thousands) and that its rate of growth t years later was P (t) = 8 + (0.5)t + (0.03)t 2 (in thousands per year). What was its population in 1990? y 0.6 0.4 65. Figure 5.5.11 shows the graph of f (x) = 1/x on the interval [1, 2], the line joining its endpoints (1, 1) and (2, 12 ), and its tangent line at the point ( 32 , 23 ). Use this construction to estimate the value of the integral 2 1 dx 1 x 0.2 0 0 0.2 0.4 0.6 0.8 1 x FIGURE 5.5.12 Bounding the 1 . graph of f (x) = 1 + x2 5.6 THE FUNDAMENTAL THEOREM OF CALCULUS Newton and Leibniz are generally credited with the invention of calculus in the latter part of the seventeenth century. Actually, others had earlier calculated areas essentially equivalent to integrals and tangent line slopes essentially equivalent to derivatives. The great accomplishments of Newton and Leibniz were the discovery and computational exploitation of the inverse relationship between differentiation and integration. This relationship is embodied in the fundamental theorem of calculus. One part of this theorem is the evaluation theorem of Section 5.5: To evaluate b f (x) d x, a it suffices to find an antiderivative of f on [a, b]. The other part of the fundamental theorem tells us that doing so is usually possible, at least in theory: Every continuous function has an antiderivative. 362 The Fundamental Theorem of Calculus SECTION 5.6 363 The Average Value of a Function The concept of the average value of a function is useful for the proof of the fundamental theorem and has numerous important applications in its own right. The ordinary (arithmetic) average of n given numbers a1 , a2 , . . . , an is defined to be a= 1 a1 + a2 + · · · + an = n n n ai . (1) i=1 But a function f defined on an interval generally has infinitely many values f (x), so we cannot simply divide the sum of all these values by their number to find the average value of f (x). We introduce the proper notion with a discussion of average temperature. EXAMPLE 1 Let the measured temperature T during a particular 24-h day at a certain location be given by the function T = f (t), 0 t 24 (with the 24-h clock running from t = 0 at one midnight to t = 24 at the following midnight). Thus, for example, the temperatures f (1), f (2), . . . , f (24) are recorded at 1-h intervals during the day. We might define the average temperature T for the day as the (ordinary arithmetic) average of the hourly temperatures: T = 24 1 24 f (ti ), i=1 where ti = i. If we divided the day into n equal subintervals rather than into 24 1-h intervals, we would obtain the more general average n 1 T = n f (ti ). i=1 The larger n is, the closer would we expect T to be to the “true” average temperature for the entire day. It is therefore plausible to define the true average temperature by letting n increase without bound. This gives n 1 n T = lim n→∞ f (ti ). i=1 The right-hand side resembles a Riemann sum, and we can make it into a Riemann sum by introducing the factor t = b−a , n where a = 0 and b = 24. Then T = lim n→∞ = lim n→∞ 1 n · n b−a 1 b−a 1 lim = b − a n→∞ n f (ti ) · i=1 n f (ti ) · i=1 n b−a n b−a n f (ti ) t = i=1 Thus 1 T = 24 24 1 b−a f (t) dt b f (t) dt. a (2) 0 under the assumption that f is continuous, so the Riemann sums converge to the inte◗ gral as n → ∞. 363 364 CHAPTER 5 The Integral The final result in Eq. (2) is the integral of the function divided by the length of the interval. Example 1 motivates the following definition: DEFINITION Average Value of a Function Suppose that the function f is integrable on [a, b]. Then the average value y of y = f (x) for x in the interval [a, b] is b 1 y= f (x) d x. (3) b−a a We can rewrite Eq. (3) in the form b f (x) d x = y · (b − a). (4) a If f is positive-valued on [a, b], then Eq. (4) implies that the area under y = f (x) over [a, b] is equal to the area of a rectangle with base length b − a and height y (Fig. 5.6.1). y y = f (x) y a x b x FIGURE 5.6.1 A rectangle illustrating the average value y of a function. EXAMPLE 2 The average value of f (x) = x 2 for x in [0, 2] is 1 y= 2 2 1 1 3 2 4 x dx = = . x 2 3 3 0 2 0 ◗ EXAMPLE 3 In Athens, Georgia (USA) the mean daily temperature in degrees Fahrenheit t months after July 15 is closely approximated by 80 70 T = 61 + 18 cos T = f(t) T = 57 T 60 πt = f (t). 6 (5) Find the average temperature between September 15 (t = 2) and December 15 (t = 5). 50 Solution Equation (3) gives 40 2 3 4 t FIGURE 5.6.2 The temperature function T = f (t) of Example 3. 5 5 πt 1 dt 61 + 18 cos T = 5−2 2 6 1 6 · 18 πt 5 = ≈ 57◦ F. 61t + sin 3 π 6 2 Figure 5.6.2 shows the graphs of T = f (t) and T ≡ 57. Can you see that Eq. (4) ◗ implies that the two almost-triangular regions in the figure have equal areas? 364 The Fundamental Theorem of Calculus SECTION 5.6 365 Theorem 1 tells us that every continuous function on a closed interval attains its average value at some point of the interval. THEOREM 1 Average Value Theorem If f is continuous on [a, b], then b 1 f (x) d x f (x) = b−a a for some number x in [a, b]. (6) Proof Let m = f (c) be the minimum value of f (x) on [a, b ] and let M = f (d) be its maximum value there. Then, by the comparison property of Section 5.5, m = f (c) y = 1 b−a b f (x) d x f (d) = M. a Because f is continuous, we can now apply the intermediate value property. The number y is between the two values m and M of f , and consequently, y itself must be a value of f . Specifically, y = f (x) for some number x between a and b. This yields Eq. (6). ◆ REMARK Whereas y denotes the average value of the function y = f (x), the point x where this average value is attained is not, in general, itself an average value of x. EXAMPLE 4 If v(t) denotes the velocity function of a sports car accelerating during the time interval a t b, then the car’s average velocity is given by 1 v= b−a b v(t) dt. a The average value theorem implies that v = v t for some number t in [a, b]. Thus t is an instant at which the car’s instantaneous velocity is equal to its average velocity ◗ over the entire time interval. The Fundamental Theorem We state the fundamental theorem of calculus in two parts. The first part is the fact that every function f that is continuous on an interval I has an antiderivative on I . In particular, an antiderivative of f can be obtained by integrating f in a certain way. Intuitively, in the case f (x) > 0, we let F(x) denote the area under the graph of f from a fixed point a of I to x, a point of I with x > a. We shall prove that F (x) = f (x). We show the construction of the function F in Fig. 5.6.3. More precisely, we define the function F as follows: y y = f(x) f(t) F(x) = F(x) a x t x+h x FIGURE 5.6.3 The area function F is an antiderivative of f . x f (t) dt, a where we use the dummy variable t in the integrand to avoid confusion with the upper limit x. The proof that F (x) = f (x) will be independent of the supposition that x > a. 365 366 CHAPTER 5 The Integral THE FUNDAMENTAL THEOREM OF CALCULUS Suppose that f is continuous on the closed interval [a, b]. Part 1: If the function F is defined on [a, b] by x F(x) = f (t) dt, (7) a then F is an antiderivative of f . That is, F (x) = f (x) for x in [a, b]. Part 2: If G is any antiderivative of f on [a, b], then b b f (x) d x = G(x) = G(b) − G(a). a (8) a Proof of Part 1 By the definition of the derivative, x+h x F(x + h) − F(x) 1 f (t) dt − f (t) dt . = lim F (x) = lim h→0 h→0 h h a a But x+h f (t) dt = a x x+h f (t) dt + a f (t) dt x by the interval union property of Section 5.5. Thus 1 x+h f (t) dt. F (x) = lim h→0 h x The average value theorem tells us that 1 x+h f (t) dt = f t h x x t x+h Flange Bead Washer Wire FIGURE 5.6.4 The bead at t trapped between the washer at x + h and the flange at x. for some number t in [x, x + h]. Finally, we note that t → x as h → 0. Thus, because f is continuous, we see that 1 x+h f (t) dt = lim f t = lim f t = f (x). F (x) = lim h→0 h x h→0 t→x Hence the function F in Eq. (7) is, indeed, an antiderivative of f . ◆ REMARK Figure 5.6.4 indicates why t must approach x as h → 0. As the moving washer at x + h approaches the fixed flange at x, the bead t between them has nowhere else to go. Proof of Part 2 Here we apply Part 1 to give a proof of the evaluation theorem in Section 5.5. If G is any antiderivative of f , then—because it and the function F of Part 1 are both antiderivatives of f on the interval [a, b]—we know that G(x) = F(x) + C on [a, b] for some constant C. To evaluate C, we substitute x = a and obtain C = G(a) − F(a) = G(a), because F(a) = a f (t) dt = 0. a Hence G(x) = F(x) + G(a). In other words, F(x) = G(x) − G(a) 366 The Fundamental Theorem of Calculus SECTION 5.6 for all x in [a, b]. With x = b this gives G(b) − G(a) = F(b) = b 367 f (x) d x, a ◆ which establishes Eq. (8). Sometimes the fundamental theorem of calculus is interpreted to mean that differentiation and integration are inverse processes. Part 1 can be written in the form d dx x f (t) dt = f (x) (9) a if f is continuous on an open interval containing a and x. That is, if we first integrate the function f (with variable upper limit of integration x) and then differentiate with respect to x, the result is the function f again. So differentiation “cancels” the effect of integration of continuous functions. Moreover, Part 2 of the fundamental theorem can be written in the form x (10) G (t) dt = G(x) − G(a) a if we assume that G is continuous. If so, this equation means that if we first differentiate the function G and then integrate the result from a to x, the result can differ from the original function G by, at worst, the constant G(a). This means that integration “cancels” the effect of differentiation when a is chosen so that G(a) = 0. 2 Computational Applications 1 Examples 1 through 4 of Section 5.5 illustrate the use of Part 2 of the fundamental theorem in the evaluation of integrals. Additional examples appear in the end-of-section problems, in this section, and in Section 5.7. Example 5 illustrates the necessity of splitting an integral into a sum of integrals when its integrand has different antiderivative formulas on different intervals. R y 0 −1 y = f(x) −2 −2 −1 0 x 1 2 FIGURE 5.6.5 The region of Example 5. Find the area A of the region R bounded above by the graph of y = f (x) and below by the x-axis. Solution The x-intercepts shown in the figure are x = −1 (where 1 − x 2 = 0 and x < 0) and x = π/2 (where cos x = 0 and x > 0). Hence 0 π/2 π/2 f (x) d x = (1 − x 2 ) d x + cos x d x A= y = x3 − x2 − 6x 8 −1 4 y 0 EXAMPLE 5 Figure 5.6.5 shows the graph of the function f defined by cos x if x 0, f (x) = 1 − x 2 if x 0. −2 1 = x − x3 3 R1 3 R2 −4 −1 −1 + sin x π/2 0 0 5 2 = +1= . 3 3 ◗ EXAMPLE 6 Figure 5.6.6 shows the graph of −8 −4 0 f (x) = x 3 − x 2 − 6x. −2 0 x 2 FIGURE 5.6.6 The graph y = x 3 − x 2 − 6x of Example 6. 4 Find the area A of the entire region R bounded by the graph of f and the x-axis. Solution The region R consists of the two regions R1 and R2 and extends from x = −2 to x = 3. The area of R1 is 0 0 3 2 2 1 4 1 3 A1 = (x − x − 6x) d x = 4 x − 3 x − 3x = 16 . 3 −2 −2 367 368 CHAPTER 5 The Integral But on the interval (0, 3), the function f (x) is negative-valued, so to get the (positive) area A2 of R2 , we must integrate the negative of f : 3 3 3 2 2 1 4 1 3 (−x + x + 6x) d x = − 4 x + 3 x + 3x = 63 . A2 = 4 8 0 4 R1 y 0 −2 A = A1 + A 2 = 3 y = ⎜x3 − x2 − 6x ⎜ −4 −8 −4 −2 0 x 0 Consequently the area of the entire region R is R2 2 FIGURE 5.6.7 The graph y = |x 3 − x 2 − 6x| of Example 6. + 16 3 63 4 = 253 12 ≈ 21.08. In effect, we have integrated the absolute value of f (x): 3 | f (x)| d x A= −2 0 4 = −2 3 (x 3 − x 2 − 6x) d x + (−x 3 + x 2 + 6x) d x = 0 253 . 12 Compare the graph of y = | f (x)| in Fig. 5.6.7 with that of y = f (x) in Fig. 5.6.6. EXAMPLE 7 Evaluate 2 −1 ◗ |x 3 − x| d x. Solution We note that x 3 − x 0 on [−1, 0], that x 3 − x 0 on [0, 1], and that x 3 − x 0 on [1, 2]. So we write 0 1 2 2 |x 3 − x| d x = (x 3 − x) d x + (x − x 3 ) d x + (x 3 − x) d x −1 −1 = = − 12 x 2 1 4 x 4 1 4 + 0 0 1 4 1 + −1 1 2 x 2 + 2 − − 14 = − 14 x 4 11 4 + 0 1 1 4 x 4 2 − 12 x 2 = 2.75. 1 ◗ Part 1 of the fundamental theorem of calculus says that the derivative of an integral with respect to its upper limit is equal to the value of the integrand at the upper limit. For example, if x y(x) = t 3 sin t dt, 0 then dy = x 3 sin x. dx Example 8 is a bit more complicated in that the upper limit of the integral is a nontrivial function of the independent variable. EXAMPLE 8 Find h (x) given h(x) = x2 t 3 sin t dt. 0 Solution Let y = h(x) and u = x 2 . Then u y= t 3 sin t dt, 0 so dy = u 3 sin u du by the fundamental theorem of calculus. Then the chain rule yields h (x) = 368 dy dy du = · = (u 3 sin u)(2x) = 2x 7 sin x 2 . dx du d x ◗ The Fundamental Theorem of Calculus SECTION 5.6 369 Initial Value Problems Note that if y(x) = x f (t) dt, (11) a then y(a) = 0. Hence y(x) is a solution of the initial value problem dy = f (x), y(a) = 0. dx To get a solution of the initial value problem dy = f (x) dx (12) y(a) = b, (13) f (t) dt. (14) we need only add the desired initial value: x y(x) = b + a EXAMPLE 9 Express as an integral the solution of the initial value problem dy = sec x, dx y(2) = 3. Solution With a = 2 and b = 3, Eq. (14) gives x sec t dt. y(x) = 3 + (15) (16) 2 With our present knowledge, we cannot antidifferentiate sec t, but for a particular value of x the integral in Eq. (16) can be approximated using Riemann sums. For instance, with x = 4 a calculator with an INTEGRATE key gives 4 sec t dt ≈ −2.5121. 2 Hence the value of the solution in Eq. (16) at x = 4 is y(4) ≈ 3 − 2.5121 = 0.4879. ◗ 5.6 TRUE/FALSE STUDY GUIDE Use the following true/false items to check your reading and review of this section. You may consult the hints provided in the answer section. 1. If f is integrable on [a, b], then its average value there is b 1 y= f (x) d x. b−a a 2. The average value of f (x) = x 2 on [0, 2] is 83 . 3. If f is continuous on [a, b] with average value y there, then y = f (x) for some number x in [a, b]. 4. If v(t) denotes the velocity—assumed continuous—of a car traveling in a straight line during the time interval a t b, then the average velocity of the car over that interval is b 1 v= v(t) dt. b−a a 5. The fundamental theorem of calculus implies that if f is continuous on [a, b], then f has an antiderivative there. 369 370 CHAPTER 5 The Integral 6. If f is continuous on [a, b] and F(x) = x f (t) dt, a then F (x) = f (x). 7. The fundamental theorem of calculus implies that if f is continuous on [a, b] and G = f there, then b f (x) d x = G(b) − G(a). a 8. The area bounded by the graph of f (x) = x 3 − x 2 − 6x and the x-axis is 2 11 . |x 3 − x| d x = 9. 4 −1 x2 10. If h(x) = t 3 sin t dt, then h (x) = x 3 sin x. 253 . 12 0 5.6 CONCEPTS: QUESTIONS AND DISCUSSION 1. Does every function defined on a closed interval have an average value there? 2. Suppose that the function f defined on the closed interval [a, b] has an average value there. Does f necessarily attain this average value there? Either show that it does or give an example showing it need not. 3. Discuss the validity of the assertion that the fundamental theorem of calculus simply says that “every function is the derivative of its integral and the integral of its derivative.” Does the fundamental theorem of calculus say this? Does it say more than this? 4. Suppose that a particle moving along the x-axis has position x = f (t) and velocity v = f (t) at time t. Interpret the values of the integrals b b f (t) dt and | f (t)| dt a a in terms of the change in position of the particle and the distance it travels. What’s the difference? 5.6 PROBLEMS In Problems 1 through 12, find the average value of the given function on the specified interval. 1. f (x) = x 4 ; [0, 2] √ 2. g(x) = x ; [1, 4] √ 3. h(x) = 3x 2 x 3 + 1; 4. f (x) = 8x; [0, 4] 5. g(x) = 8x; [−4, 4] 6. h(x) = x 2 ; [−4, 4] 7. f (x) = x ; 3 −1 (x 3 + 2)2 d x −1 18. −2 2 20. −1 x −x +3 dx √ 3 x 2 |x| d x e2x−1 d x 10. g(x) = sin 2x; 12. g(t) = e2t ; [0, π/2] [−1, 1] 2 24. √ x − x d x 6 π 0 2 1 0 1 23. −2 8 27. 4 11 1 dx x −1 1 2 t− dt 2t e2x − 1 dx ex 2 25. 0 dx √ 9x 3 3t − 5 dt t4 sin x cos x d x π/3 sin 3x d x 28. 1 21. 0 26. 3 19. 0 1 17. 1 22. Evaluate the integrals in Problems 13 through 28. 3 d x (Here d x stands for 1 d x.) 13. 370 1 −1 8. g(x) = x −1/2 ; [1, 4] √ 9. f (x) = x + 1 ; [0, 3] [0, π] 4 15. 1 [0, 5] 11. f (x) = sin 2x; (y 5 − 1) dy 16. [0, 2] 2 14. |x 2 − 1| d x 1 dx x The Fundamental Theorem of Calculus SECTION 5.6 In Problems 29 through 32, the graph of f and the x-axis divide the x y-plane into several regions, some of which are bounded. Find the total area of the bounded regions in each problem. 1 29. f (x) = 1 − x 4 if x 0; (Fig. 5.6.8) x 30. f (x) = (π/2) sin x on [0, π/2]; [π/2, π ] (Fig. 5.6.9) f (x) = x(π − x) on 3 y = 1 − x3 y = 1 − x4 y = (π /2)2 sin x FIGURE 5.6.12 The sphere of Problem 38. 2 y = x(π − x) 1 y 0 y 39. Figure 5.6.13 shows a cross section at distance y from the vertex of a cone with base radius 1 and height 2. Find the average area of this cross section for 0 y 2. 0 −1 −1 −1 0 x r f (x) = 1 − x 3 if x 0 2 1 371 −2 1 0 1 2 3 x FIGURE 5.6.8 Problem 29. FIGURE 5.6.9 Problem 30. y 2 31. f (x) = x 3 − 9x 32. f (x) = x 3 − 2x 2 − 15x (Fig. 5.6.11) y = x3 − 9x 10 r (Fig. 5.6.10) 1 y = x3 − 2x2 − 15x 40 20 y 0 y 0 −20 −10 −4 − 40 −2 0 x 2 4 −4 0 x 4 FIGURE 5.6.10 Problem 31. FIGURE 5.6.11 Problem 32. 33. Rosanne drops a ball from a height of 400 ft. Find the ball’s average height and its average velocity between the time it is dropped and the time it strikes the ground. FIGURE 5.6.13 The cone of Problem 39. 40. A sports car starts from rest (x = 0, t = 0) and experiences constant acceleration x (t) = a for T seconds. Find, in terms of a and T , (a) its final and average velocities and (b) its final and average positions. 41. (a) Figure 5.6.14 shows a triangle inscribed in the region that lies between the x-axis and the curve y = 9−x 2 . Express the area of this triangle as a function A(x) of the x-coordinate of its upper vertex P. (b) Find the average area A of A(x) for x in the interval [−3, 3]. (c) Sketch a triangle as in Fig. 5.6.14 that has the area A found in part (b). How many different such triangles are there? 34. Find the average value of the animal population P(t) = 100 + 10t + (0.02)t 2 over the time interval [0, 10]. 35. Suppose that a 5000-L water tank takes 10 min to drain and that after t minutes, the amount of water remaining in the tank is V (t) = 50(10 − t)2 liters. What is the average amount of water in the tank during the time it drains? 36. On a certain day the temperature t hours past midnight was π T (t) = 80 + 10 sin (t − 10) . 12 y P(x, y) y = 9 − x2 x FIGURE 5.6.14 The typical triangle of Problem 41. What was the average temperature between noon and 6 P. M .? 37. Suppose that a heated rod lies along the interval 0 x 10. If the temperature at points of the rod is given by T (x) = 4x(10 − x), what is the rod’s average temperature? 38. Figure 5.6.12 shows a cross section at distance x from the center of a sphere of radius 1. Find the average area of the cross section for 0 x 1. 42. (a) Figure 5.6.15 shows a rectangle inscribed in the firstquadrant region that lies between the x-axis and the line y = 10 − x. Express the area of this rectangle as a function A(x) of the x-coordinate of its vertex P on the line. (b) Find the average area A of A(x) for x in the interval [0, 10]. (c) Sketch a rectangle as in Fig. 5.6.15 that has the area A found in part (b). How many different such rectangles are there? 371 372 CHAPTER 5 The Integral y sin x 56. f (x) = ! 1 − t 2 dt sin x (t + 1) dt 58. f (x) = 2 59. f (x) = 3 1 P(x, y) sin t dt 0 0 x + y = 10 x2 57. f (x) = x 2 +1 1 dt t ex ln(1 + t 2 ) dt 60. f (x) = 1 x Use integrals (as in Example 9) to solve the initial value problems in Problems 61 through 64. 61. FIGURE 5.6.15 The typical rectangle of Problem 42. 1 dy = , dx x y(1) = 0 dy 1 π = , y(1) = 2 dx 1+x 4 √ dy 63. = 1 + x 2 , y(5) = 10 dx 62. 43. (a) Figure 5.6.16 shows a rectangle inscribed in the semicircular √ region that lies between the x-axis and the graph y = 16 − x 2 . Express the area of the rectangle as a function A(x) of the x-coordinate of its vertex P on the line. (b) Find the average area A of A(x) for x in the interval [0, 4]. (c) Sketch a rectangle as in Fig. 5.6.16 that has the area A found in part (b). How many different such rectangles are there? y 64. dy = tan x, dx y(1) = 2 65. The fundamental theorem of calculus seems to say that 1 −1 y = 16 − x2 dx 1 1 = − = −2, x2 x −1 P(x, y) in apparent contradiction to the fact that 1/x 2 is always positive. What’s wrong here? x 66. Prove that the average rate of change f (b) − f (a) b−a FIGURE 5.6.16 The typical rectangle of Problem 43. 44. Repeat Problem 43 in the case that the rectangle has two vertices on the x-axis and two on the√ parabola y = 16 − x 2 (rather than on the semicircle y = 16 − x 2 ). You may need to use a calculator or computer to find the base of a rectangle whose area is the average A of A(x) for x in [0, 4]. In Problems 45 through 49, apply the fundamental theorem of calculus to find the derivative of the given function. x t ! 45. f (x) = (t 2 + 1)17 dt 46. g(t) = x 2 + 25 d x −1 0 z x √ 1 3 47. h(z) = u − 1 du 48. A(x) = dt t 2 1 10 (et − e−t ) dt 49. f (x) = of the differentiable function f on [a, b] is equal to the average value of its derivative on [a, b]. 67. The graph y = f (x), 0 x 10 is shown in Fig. 5.6.17. Let x f (t) dt. g(x) = 0 (a) Find the values g(0), g(2), g(4), g(6), g(8), and g(10). (b) Find the intervals on which g(x) is increasing and those on which it is decreasing. (c) Find the global maximum and minimum values of g(x) for 0 x 10. (d) Sketch a rough graph of y = g(x). x In Problems 50 through 53, G(x) is the integral of the given function f (t) over the specified interval of the form [a, x], x > a. Apply Part 1 of the fundamental theorem of calculus to find G (x). √ t 50. f (t) = 2 ; [2, x] 51. f (t) = t + 4 ; [0, x] t +1 √ 53. f (t) = t 3 + 1 ; [1, x] 52. f (t) = sin3 t; [0, x] In Problems 54 through 60, differentiate the function by first writing f (x) in the form g(u), where u denotes the upper limit of integration. x2 ! 3x 54. f (x) = 1 + t 3 dt 55. f (x) = sin t 2 dt 0 372 2 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 y = f(x) 0 2 4 6 8 x FIGURE 5.6.17 Problem 67. 10 Integration by Substitution SECTION 5.7 68. Repeat Problem 67, except use the graph of the function f shown in Fig. 5.6.18. 373 10 5 5 4 3 2 1 y 0 −1 −2 −3 −4 −5 y 0 -5 y = f(x) y = x sin x -10 0 2 4 6 x 8 10 12 FIGURE 5.6.19 Problem 69. 0 2 4 6 8 70. Repeat Problem 69, except use the function 10 x f (x) = FIGURE 5.6.18 Problem 68. sin x x on the interval [0, 4π ] (as shown in Fig. 5.6.20). Take f (0) = 1 because (sin x)/x → 1 as x → 0. 69. Figure 5.6.19 shows the graph of the function f (x) = x sin x on the interval [0, 4π ]. Let g(x) = 1 0.8 0.6 0.4 0.2 y 0 − 0.2 − 0.4 − 0.6 − 0.8 −1 x f (t) dt. 0 (a) Find the values of x at which g(x) has local maximum and minimum values on the interval [0, 4π]. (b) Where does g(x) attain its global maximum and minimum values on [0, 4π ]? (c) Which points on the graph y = f (x) correspond to inflection points on the graph y = g(x)? (d) Sketch a rough graph of y = g(x). y = sin x x 0 2 4 6 x 8 10 12 FIGURE 5.6.20 Problem 70. 5.7 INTEGRATION BY SUBSTITUTION The fundamental theorem of calculus in the form b b f (x) d x = f (x) d x a (1) a implies that we can readily evaluate the definite integral on the left if we can find the indefinite integral (that is, antiderivative) on the right. We now discuss a powerful method of antidifferentiation that amounts to “the chain rule in reverse.” This method is a generalization of the “generalized power rule in reverse,” u n du = u n+1 +C n+1 (n = −1), which we introduced in Section 5.2. Equation (2) is an abbreviation for the formula [g(x)]n+1 +C [g(x)]n g (x) d x = n+1 (n = −1) (2) (3) that results when we write u = g(x), du = g (x) d x. So to apply Eq. (2) to a given integral, we must be able to visualize the integrand as a product of a power of a differentiable function g(x) and its derivative g (x). 373 374 CHAPTER 5 The Integral EXAMPLE 1 With u = 2x + 1, du = 2 d x, we see that 1 u6 + C = (2x + 1)6 + C. (2x + 1)5 · 2 d x = u 5 du = 6 6 EXAMPLE 2 ! 2 2x 1 + x d x = (1 + x 2 )1/2 · 2x d x (a) (u = 1 + x 2 , = u 1/2 du = u 3/2 3 2 ◗ du = 2x d x) + C = 23 (1 + x 2 )3/2 + C. (b) Similarly, but with u = 1 + e x and du = e x d x, we get √ 1 ex dx = √ √ du = 2 u + C u 1 + ex √ = 2 1 + e x + C. ◗ Equation (3) is the special case f (u) = u n of the general integral formula f (g(x)) · g (x) d x = f (u) du. (4) The right-hand side of Eq. (4) results when we make the formal substitutions u = g(x), du = g (x) d x on the left-hand side. One of the beauties of differential notation is that Eq. (4) is not only plausible but is, in fact, true—with the understanding that u is to be replaced with g(x) after the indefinite integration on the right-hand side of Eq. (4) has been carried out. Indeed, Eq. (4) is merely an indefinite integral version of the chain rule. For if F (x) = f (x), then Dx F(g(x)) = F (g(x)) · g (x) = f (g(x)) · g (x) by the chain rule, so F (g(x)) · g (x) d x = F(g(x)) + C f (g(x)) · g (x) d x = = F(u) + C [u = g(x)] = f (u) du. Equation (4) is the basis for the powerful technique of indefinite integration by substitution. It may be used whenever the integrand function is recognized to be of the form f (g(x)) · g (x). EXAMPLE 3 Find 374 ! x 2 x 3 + 9 d x. Integration by Substitution SECTION 5.7 375 Solution Note that x 2 is, to within a constant factor, the derivative of x 3 + 9. We can, therefore, substitute u = x 3 + 9, du = 3x 2 d x. (5) The constant factor 3 can be supplied if we compensate by multiplying the integral by 1 . This gives 3 ! 1 1 (x 3 + 9)1/2 · 3x 2 d x = u 1/2 du x2 x3 + 9 dx = 3 3 1 u 3/2 2 2 = · 3 + C = u 3/2 + C = (x 3 + 9)3/2 + C. 3 9 9 2 An alternative way to carry out the substitution in (5) is to solve du = 3x 2 d x and then write for x2 dx = 1 3 du = 1 3 (x 3 + 9)1/2 d x = du, u 1/2 · 1 3 u 1/2 du, concluding the computation as before. ◗ The following three steps in the solution of Example 3 are worth special mention: • • • The differential d x along with the rest of the integrand is “transformed,” or replaced, in terms of u and du. Once the integration has been performed, the constant C of integration is added. A final resubstitution is necessary to write the answer in terms of the original variable x. Substitution in Trigonometric and Exponential Integrals By now we know that every differentiation formula yields—upon “reversal”—a corresponding a