SoylentNews is people
SoylentNews
Hundreds of Springer books for free download
The publisher Springer has made 473 of their books free to download during these covid shutdown times. I came across this as I was purusing the R-Bloggers site and I saw a package announcement that lets you download them very conveniently. Apparently that package was based upon this python project.
Lots of good math, physics, compsci, etc.
Where has all the ECC RAM gone?
It's about time I built a new PC. The old Phenom II X6 1045T is getting a bit long in the tooth.
I was looking at some sort of AMD X570 chip set motherboard from ASUS and a Ryzen CPU and I was looking for RAM to go with it. I have ECC RAM in my Phenom II and it has detected and fixed the odd error here and there over the 10 years I've had it so I want to put ECC RAM in my new machine.
It seems that it's as rare as hen's teeth these days. In days of yore you just went to www.crucial.com and put in how much you want and it would spout a list of all the options. Now there's none. I believe I would need two sticks of unbuffered DDR4 with ECC.
Some re-sellers have it listed but you have to go looking for it under "Server Memory" and it's mixed up with all kinds of weird stuff. Most of the stuff for sale seems to be very high clock rate, unbuffered and with no error checking to make the games run 2% faster.
What is the world coming to?
Boostrapping Rudimentary AI 2: Discovering the Obvious
Well, I've procrastinated quite a bit from my previous journal, yet despite that, I've managed to come across a couple of interesting, though rather obvious things. While looking for inspiration, I ran across a paper with the lengthy title, Leveraging asynchronous parallel computing to produce simple genetic programming computational models. The idea was to run a genetic algorithm (GA) on a parallel computation system completely asynchronously, by generating new programs from a pool of existing programs, and putting the more successful ones back into the pool as they are generated.
The key property is that successful programs that complete faster can return to the pool quicker (presuming of course, that you have a lot of processors/threads generating these programs rather than just one) and have the potential to evolve faster than longer running programs. So in theory, one doesn't need to include explicit time constraints in their criteria for successful programs because there's this built-in bias towards faster running programs (as long as they are successful enough to stay in the pool). The gist of the paper is that this is indeed the case in practice as theory (tested on eight different problems IIRC).
Among other things, this is in large part how real world evolution works, particularly at the microbe level. At the macroscopic level, you get some survival filters, particularly seasonal stuff, that shapes things like when organisms breed and how long they live.
From my provincial point of view, it was eye-opening because I hadn't given a great deal of thought as to how to structure the genetic algorithms that I'll attempt to start my bootstrapping scheme. The old approach that I did many years ago, the "Generational GA", was to generate a large pool of "genes", pick the best 5-10% of them, put that group in a new pool, generate from them a new pool (next generation), and start over. Now, though I'm not doing any parallelism, I can still greatly reduce the memory load by just dealing strictly with the "best" pool from the start. Generating a new gene from two in the pool, the "Incremental GA", and if the new one is better than the worst gene, replace the worst. In addition, it appears to be a little faster in converging to a useful solution (though I haven't quantified it).
The Lattice Coloring Problem
Finally, let me briefly describe what I'm presently working on. I start with a 60 by 60 grid (the number 60 is chosen because it allows for patterns with periods up to 6) of "colorings", every vertex is numbered between 0 and N-1 (I presently use N=4). A small patch of the grid looks something like:
0 3 2 3 1
1 3 3 1 1
2 0 0 3 1
3 0 2 2 1
The energy of this lattice is based on a randomized function that takes four adjacent vertices of a 2 by 2 box and assigns a real number to them.
w x
y z
Then sum over all such boxes of the grid, wrapping around both top and bottom, left and right. For the 60 by 60 grid, you end up doing 3600 such sums.
The problem then is to find a color pattern that minimizes this total energy. One gets genuine random colorings as the lowest energy state only for rare cases (such as when all possible colorings have the same energy). Instead interesting patterns appear.
Some common examples (just providing the pattern, not the energy function that these patterns happen to be minimum energy for):
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
Solid color is the most common.
1 2 1 2
1 2 1 2
1 2 1 2
1 2 1 2
Alternating solid rows or solid columns is next.
0 3 0 3
3 0 3 0
0 3 0 3
3 0 3 0
Checkerboard.
2 3 2 2 2 2
2 2 2 3 2 3
2 3 2 2 2 2
2 2 2 3 2 3
A pattern with a bit of randomness thrown in.
On the last one, the building block is a 2 by 2 block of 2s with one 3. You can shift either the rows or the columns (but not both), randomly offset by 1 (here, it's columns), and still have a minimum energy. I've looked a little bit for a genuine nonperiodic pattern (which would be a quasicrystal), but haven't found it yet.
Here's an example of a four color coloring that can be a lowest energy state:
0 0 0 0
1 1 1 1
2 2 2 2
3 3 3 3
The nice thing about this particular problem is that it's very easy to turn into a genetic algorithm. I basically just copy/paste random rectangles from one grid to another. I can also mutate by copy/pasting from a randomly generated grid to a good grid. It's one of the simpler problems one can apply the GA approach to.
So how fast does it converge? Not very. But starting with a pool of ten randomly generated grids, I usually can come up with a reasonable try after a few hundred to few thousand iterations. What I've discovered is that convergence is fastest when the lowest energy pattern isn't near other patterns in terms of energy. For example, if all 2s is almost as low an energy as all 1s, we would see the 0 and 3 colors rapidly disappear, but then it'd take longer for the 2s to vanish. Sometimes the program would halt on all 2s or even not get to a state of all 1s or all 2s.
I've also had difficulty reaching complex pattern lowest energy states. They tend to converge slower than for the simpler color patterns mentioned above. And if the pattern has a period that doesn't divide evenly into 60 (such as 7 or 9), the pattern will always be truncated. That can introduce higher energy defects which take a while to settle down.
As a final remark, I could look at different numbers of colors or different sizes of grid, as well as figuring out how fast these things converge under different circumstances. I haven't bothered at this time to do so.
What's Next?
Presently, I'm still avoiding actual genetic programs and am instead looking at fitting a table of data (each column corresponding to a parameter) to a polynomial function of the parameters. It's still a pretty easy problem for GA to attack, but involves generation of arbitrary monomials which involves similar tree-building to what generating programs would involve. I presently am thinking of using it to generate a differential equation for a vector differential equation I messed with for a while. Mathematica doesn't seem effective at dealing with it in its present form. Maybe, if I convert the problem to a bunch of separate ODEs, it'll work better.
I'm also thinking about making a default mutation algorithm of breeding a gene with a randomly generated gene since I need both subalgorithms anyway for my basic GA system.
Opinion: the meaning of TDS
In my opinion, the term TDS has exactly one meaning: "I have no actual rebuttal to what was said."
It is (IMO falsely) claimed that TDS is used when someone trivially overreacts to any tiny thing that Trump says or does. However in my experience, I've never seen it used this way.
The way TDS is typically used, at least on SN, is when someone makes a point, in many cases supported at least minimally with facts and/or references, critical of the president, and the TDS-dropper has no actual rebuttal.
The correct reaction to seeing a TDS-bomb? Stop. You've lost me right there. I'm done reading. If you had an actual point to make you would have already made it.
Labeling valid criticism as deranged is itself, IMO, deranged.
Edit, additional thought:
Furthermore, throwing out TDS is not intended to contribute or rebut in any meaningful way, it is intended to be hurtful. It serves NO OTHER purpose. Like throwing a grenade, it has no cost to the thrower.
Science v. Politics
Jim Al-Khalili has written in the Guardian an article entitled Doubt is essential for science - but for politicians, it's a sign of weakness.
In the article he addresses the phenomenon of social media echo chambers, politics, news reporting, cognitive dissonance and the way scientific progress is made. Doubt is essential to science and the Scientific Method. Scientists ask questions, form theories, perform experiments and refine the theories in the light of new evidence.
When the public hears that new scientific evidence has informed a sudden change in government policy, the tendency is to conclude that the scientists don’t know what they’re doing, and therefore can’t be trusted. It doesn’t help that politicians are remarkably bad at communicating scientific information clearly and transparently, while journalists are often more adept at asking questions of politicians than they are of scientists.
And of conspiracy theorists:
Often, in the case of such ideological beliefs, we hear the term “cognitive dissonance”, whereby someone feels genuine mental discomfort when confronted with evidence that contradicts a view they hold. This can work to reinforce pre-existing beliefs. Ask a conspiracy theorist this: what would it take for them to change their minds? Their answer, because they are so utterly committed to their view, is likely to be that nothing would. In science, however, we learn to admit our mistakes and to change our minds to account for new evidence about the world.
In another Guardian article BBC's Covid-19 reporters: 'I wanted to show the reality but was deeply troubled by what I saw’, BBC reporters Fergus Walsh and Hugh Pym were interviewed about their reporting on the COVID-19 crisis from the NHS front line hospitals treating critically ill patients.
They are trying to report the facts so that good questions can be asked of the government. Walsh says, "I feel a huge sense of responsibility to get the tone right, the messaging right, and to show people the reality of what’s going on in our hospitals.”
“Given the technology [thanks to social distancing, journalists now attend remotely] the press conferences have been tricky,” says Pym. “People say: why didn’t you follow up? But until they started allowing follow-up questions, that was difficult. The government side has offered us the chance to do filming on testing or drug trials, but it’s so fast-moving: the communications team, civil servants and ministers are under huge pressure.”
Trump, Disinfectants, and the Media Spin [Viva Frei]
<teaser>
We are either living in a post-truth world or a truthless world and either way it is fricking depressing.
<intro music>
Viva Frei Montreal litigator turned Youtuber it is yet another car vlawg it is going to be yet another toned down car vlawg because I yet again have a kid sleeping in the backseat.
So it's Friday, there's news, which means that there is fake news and I think I'm gonna go with the term Fake News Friday.
<intro music>
In todays episode of Fake News Friday, it seems the Media would have us believe that Donald Trump actually seriously suggested injecting bleach or disinfectant in order to kill the MySharonCyrus,™ and in case you don't believe me let me just read you a few headlines:
</teaser>
<link>
https://youtu.be/TbqGop2cfVc?t=37
</link>
<me>
Any mistakes in my transcription are my own, if there's a non skynet url please post it I'll mod it right up. I'm really not going to sit here and transcribe the entire thing.
But I love that he doesn't waste our time, he speaks expeditiously and gets to the point without too many meandering digressions.
Trump is a gigantic douchebag. But I have no trouble empathizing with people that call him "God Emperor." It's all in what he's running /against./
Anti-Trump people, listen to me. When you make shit up, or uncritically repeat someone who makes shit up, and it turns out to be bullshit, you don't make us look good, you make him look less bad by comparison. You make us look like lunatics.
CUT THIS SHIT OUT!
And pro-Trump people, assuming he had good intentions at all, clearly he's been stymied. The next four years are going to be particularly dangerous, in terms of whatever benefits he bought being reversed harshly. Do NOT give him unconditional support, well, too late for that to matter now I suspect but it's still true. Beyond that though, look at congressional districts. Don't vote based on party, vote based on policy. If he truly is against the forever war, then the only explanation is he doesn't have enough support in Congress - and you don't give him more support by blindly voting (R). You give him more support on this issue by voting for the policy, not the party. There are a few (R)s a few (D)s a few (L)s a few (G)s and I'm sure a few more I forgot to mention, but look for people that advocate sensible policy whenever there is anyone that comes reasonably close to that.
And when no one does, just vote against the incumbent.
Ok, that's enough of </me.>
True leaders lead by example
From the BBC:
Coronavirus: Outcry after Trump suggests injecting disinfectant as treatment
(article includes direct video quotation -- despite any efforts to retroactively change the transcript)
US President Donald Trump has been lambasted by the medical community after suggesting research into whether coronavirus might be treated by injecting disinfectant into the body.
He also appeared to propose irradiating patients' bodies with UV light, an idea dismissed by a doctor at the briefing.
Another of his officials had moments earlier said sunlight and disinfectant were known to kill the infection.
Disinfectants are hazardous substances and can be poisonous if ingested.
Even external exposure can be dangerous to the skin, eyes and respiratory system.
True leaders lead by example. Lead by being the first one into the battle with others following. Be the first one to demonstrate the powerful efficacy of your newfound treatment. Show those silly doctors and scientists how much of a stable genius you truly are. Demonstrate your truly dizzying intellect for all to see. It is well known that verifiable personal testimony of the effectiveness of your treatment is a powerful sales pitch. Consider how such courage will inspire your followers come election time.
True followers follow their glorious leader's example. As a heartening and inspiring example of selfless devotion I would point no further than Jonestown or your local Apple store. The VP should be the first in line to demonstrate the loyalty of a true follower. Inspire all the acolytes. (leaving Pelosi in charge, I think)
<no-sarcasm>
Listening to doctors and scientists and ignoring idiotic orange clowns, their daily three ring circus, and their cheerleading lap dogs at FauxNews, is really the best advice.
</no-sarcasm>
Update:
Lysol maker warns against internal use of disinfectants after Trump comments
I don't think there is any way to spin this that Trump did not mean what we all saw and heard him say. Everyone independently understood him to mean the same thing. Then they came out defending Trump saying he was being sarcastic -- which confirms that we correctly understood his message about disinfectants.
Oil goes negative
Unprecedented moves are being made in the oil markets as the lack of demand has caused the demand for storage to exceed capacity. At this hour, various sources are quoting oil at NEGATIVE $6 and some change per barrel. Yahoo Finance can't handle this. Their quote shows a roughly $25 price decline, but continues to quote it at $0.05 despite the last price being in the teens. Nobody knows the web developers are naked until commodity prices go negative.
Update -- Yahoo Finance somehow fixed their glitch. The WTI Contract is now quoted at -$18.20 on a drop of $36.47. Other sources are stating the futures contract closed at -$37.63. This doesn't mean global oil prices are negative. It's just a widely quoted contract for delivery in Oklahoma. World oil prices at various locations are still in the 20s, and later dated contracts for US delivery are in the 20s for now.
No Discussion Here
This is the journal entry mentioned in Meta: Coronaids and Subscriptions.
Census form race/origin question
While some posters have argued that there is "no such thing" as race, we are nevertheless being asked by the government to affix ourselves with a label which seems likely to end up in a database.
Then there is this question: what is my origin? Well, I didn't originate at any other country different from the one I currently reside in. Nor did my parents. Nor did my grandparents. Out of eight great-grandparents, I have reliable information indicating that two of them did originate in a different country. The remaining ones did not, or I don't have any information one way or the other.
The census form has supplied a 16-character box in which to provide an answer, and zero instructions. For bonus points, the previous question also asks about origin, making it either redundant or contradictory.
This is why I hate paperwork in general, and surveys in particular. The creator of the question insists that I take it seriously, but this is impossible because they haven't taken it seriously themselves.
It reminds me of the fabulous test questions I would sometimes encounter in school. "True or False: Cars have four wheels." And short of being able to read the teacher's mind, there is no way to answer this. If you mark it false, they'll say "the answer was true, two in front and two in back, remember?" but if you mark it true they'll say "the answer was false, because some cars have three wheels". Or maybe they thought they would be clever and count the steering "wheel" too...
