SoylentNews is people
SoylentNews
from the all-your-cryto-are-belong-to-us dept.
Google Takes Aim at Quantum Supremacy with 72-Qubit Chip
Google's Quantum AI Lab has revealed it is testing a 72-qubit quantum processor in its bid to become the first company to demonstrate quantum supremacy.
The chip, known as Bristlecone, was revealed on Monday at the annual American Physical Society meeting in Los Angeles. Given that you only need 49 or 50 qubits to demonstrate quantum supremacy – the capability of a quantum computer to outperform the largest supercomputers on certain computational tasks – a 72-qubit processor should be more than enough to achieve such a milestone. However, the error rates of such a system must be low enough so that it to be of practical use. Moreover, the ability to test a system for quantum supremacy is confounded by the very fact that classical computers cannot be used to compare test results.
Julian Kelly, a research scientist at the Quantum AI Lab who presented Bristlecone at this week's American Physical Society meeting, described the progress of their work in a Google Research blog post. In his writeup, he characterized the new chip as a "a testbed for research into system error rates and scalability of our qubit technology, as well as applications in quantum simulation, optimization, and machine learning."
Also at BGR.
(Score: 1, Interesting) by Anonymous Coward on Wednesday March 07 2018, @04:40PM (3 children)
Why can't you use classical computers compare the test results? I'm assuming they are talking about results that have "collapsed" and are not in superposition. In which case classical computers should be able to generate the same results but take longer to do so.
Or does a 50+ qubits test take way too long for a classical computer to work stuff out?
(Score: 4, Interesting) by maxwell demon on Wednesday March 07 2018, @09:44PM
Exactly what I thought. Factorize numbers that a classical computer could not factorize in reasonable time, and the classical computer can then check in virtually no time that the product of the claimed factors indeed is the original number.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 0) by Anonymous Coward on Thursday March 08 2018, @02:01PM (1 child)
A qubit takes three real numbers to fully describe its intermediate state, but determining the joint state of n qubits requires taking the outer product (because of entanglement), so you end up with 3^n real numbers to keep track of. Assuming one uses doubles at 64 bits per real number, that means for n=50 qubits you need about 7e23 bits. That's a hell of an order for a classical computer to churn through.
(Score: 0) by Anonymous Coward on Thursday March 08 2018, @02:03PM
Ugh, can't math this early. It's 4.6e25 bits.
Of course, one could check some specific problem with a known solution, but a general verification is a massive pain in the ass.