Have Scientists Disproven Google's Quantum Supremacy Claim? (scmp.com) 35
Slashdot reader AltMachine writes: In October 2019, Google said its Sycamore processor was the first to achieve quantum supremacy by completing a task in three minutes and 20 seconds that would have taken the best classical supercomputer, IBM's Summit, 10,000 years. That claim — particularly how Google scientists arrived at the "10,000 years" conclusion — has been questioned by some researchers, but the counterclaim itself was not definitive.
Now though, in a paper to be submitted to a scientific journal for peer review, scientists at the Institute of Theoretical Physics under the Chinese Academy of Sciences said their algorithm on classical computers completed the simulation for the Sycamore quantum circuits [possibly paywalled; alternative source of the same article] "in about 15 hours using 512 graphics processing units (GPUs)" at a higher fidelity than Sycamore's. Further, the team said "if our simulation of the quantum supremacy circuits can be implemented in an upcoming exaflop supercomputer with high efficiency, in principle, the overall simulation time can be reduced to a few dozens of seconds, which is faster than Google's hardware experiments".
As China unveiled a photonic quantum computer which solved a Gaussian boson sampling problem in 200 seconds that would have taken 600 million years on classical computer, in December 2020, disproving Sycamore's claim would place China being the first country to achieve quantum supremacy.
Now though, in a paper to be submitted to a scientific journal for peer review, scientists at the Institute of Theoretical Physics under the Chinese Academy of Sciences said their algorithm on classical computers completed the simulation for the Sycamore quantum circuits [possibly paywalled; alternative source of the same article] "in about 15 hours using 512 graphics processing units (GPUs)" at a higher fidelity than Sycamore's. Further, the team said "if our simulation of the quantum supremacy circuits can be implemented in an upcoming exaflop supercomputer with high efficiency, in principle, the overall simulation time can be reduced to a few dozens of seconds, which is faster than Google's hardware experiments".
As China unveiled a photonic quantum computer which solved a Gaussian boson sampling problem in 200 seconds that would have taken 600 million years on classical computer, in December 2020, disproving Sycamore's claim would place China being the first country to achieve quantum supremacy.
Someone start the movement (Score:2)
End quantum supremacy, macro matter matters.
Re: (Score:3)
Yep.
JFK will be his VP, and Hillary Clinton will be going to jail any day now.
Right, whos for Pizza?
Re: (Score:2)
So now that China’s outlawed crypto mining (Score:3)
It sounds like they’re able to use their GPUs for practical purposes?
Yea (Score:5, Interesting)
Google's claim did seem like bullshit because to prove quantum supremacy all they have to do is factorize the product of two large prime numbers faster than GNFS instead of perform some weird esoteric specialized computational task such as sampling pseudo random quantum circuits and cross-entropy benchmarking. Which journalist would challenge them on that? It feels like they announced it for short term high media publicity knowing it would take some time for someone to say "hey, wait a second.."
Re: (Score:1)
nope, no current quantum computer can factorize faster than a digital one, setup time in tens of seconds or more is too long.
Re: (Score:2)
According to Wikipedia [wikipedia.org], the largest number ever factored by Shor's algorithm is 21. That wouldn't impress anyone much. Quantum annealers have been used to factor larger numbers (over 1 trillion), but they aren't real "quantum computers", and anyway, that number would still probably take a second or less to factor on a laptop computer.
Re: (Score:2)
The point is that quantum computers at this point are nowhere near big enough to do that. Integer factoring isn't interesting until it is hundreds of bits, and the machine in question has less than 100 bits so it can't even hold a number to be factored much less run Shor's algorithm on it.
The Google "quantum supremacy" thing is on a toy problem, because that's only how big/effective the machine was -- it was only at the point where it was seriously competitive with a datacenter/supercomputer when running ag
Re: (Score:2)
Re: Yea (Score:2)
There are zillions of contrived problems that can be framed such that a classic computer cannot solve it as fast as something by else. In this case Google chose quantum circuit simulation. That is cheating, virtually the equivalent of claiming pebble supremacy by tossing a pebble into water and claiming the interference patterns and refraction were computed faster than a classical computer. Another example lighting up an indoor photography shoot and claiming that it achieves photorealistic rendering superio
Re: (Score:2)
Google's claim did seem like bullshit because ...
IBM already debunked it right after the news broke.
Correct thecnical definition of Quantum Supremacy (Score:1)
I am not defending IBM's claims, I haven't even read the press release yet. But please be aware that "quantum supremacy" has a very precise meaning that is not what 99.999% of the people (including Slashdot users) would expect.
TL;DR: 1) "quantum supremacy" is a moving target 2) it is totally possible that IBM has achieved "quantum supremacy" as we currently define it 3) this has almost surely no practical implications yet 4) integer factorization is
Re: (Score:1)
Re: google exaggerates (Score:2)
Worse the task was quantum circuit simulation, the equivalent of claiming pebble supremacy by tossing a pebble into water and claiming the interference patterns and refraction were computed faster than a classical computer. Basically true, but useless. Although in Googles case even that turned out false.
A pissing contest between Google and China (Score:2)
Claiming to achieve something... (Score:2)
...and actually achieving it are not the same thing. Let's wait until the indeoendent peer review comes out about the chinese computer first, and my money is on that never happening. 76 photon qubits machine suddenly appears out of nowhere with no intermediate stages and no one in the quantum computer world was any the wiser? Pull the other one. Sounds like propaganda. If it was that good it would be top secret and never see the light of day with the chinese military reaping the rewards for years.
Re: Claiming to achieve something... (Score:2)
Alternate theory: every time someone claims. Quantum Supremacy, it will be disproven via increasingly complex simulations of quantum computers. This will be an arms race until it is ultimately proven that P=NP, and that quantum computers are just really good at feeding in the large number of hidden variables required for the conversion.
Quantum reality: real
Superposition: almost certainly a misunderstanding
The math that currently drives quantum the theory: very real and very important and very good at predic
Re: (Score:2)
Nope (Score:2)
Google never provided any poof for their claim besides a worthless stunt. They never proved that classical "10'000" year claim for classical computers, they just claimed it for a specific unsuitable hardware and algorithm.
There is nothing to disprove here.
Re: (Score:2)
Re: (Score:2)
Yes. pretty much. Essentially all QC "breakthrough" claims are of the same quality.
One thing I noticed is that the actors in that space seem to be getting more desperate in the last few years. Maybe 50 years of R&D with no useful results are finally catching up to them and they now try lying, because the real breakthrough must be around the corner, right?
Was the answer "42"? (Score:1)
Then I call "bullshit"!
Photonic quantum computer is a stretch (Score:4, Interesting)
The device built in China is custom built to solve only one problem, to measure samples of photons, and it does that by shooting photons through beam splitters and measuring them. The splitters donâ(TM)t move, there are no logic gates, it sure as hell isnâ(TM)t Turing complete - itâ(TM)s not a computer, itâ(TM)s an experiment. An optical experiment that they labeled as a quantum computer. Computers can be programmed. They can be re-used for other things. They have memory and variables and software. Am I missing something?
Re: (Score:2)
Technically in computer science there's no such thing as a device that only solves one problem. If you solve a problem you effectively solve all problems which are practically reducible to it. It doesn't matter what the internal machinery is, or if (in the case of quantum computing) there isn't any machinery at all; it can be an infinitely long piece of tape you mark, erase, or read symbols.
Sure the quantum device may not be *programmable*, but the whole point is to have a way to solve problems that aren'
Re: (Score:2)
Technically in computer science there's no such thing as a device that only solves one problem.
Pick that nit!! So, Mr. Smart Guy, what other computer science problem can be solved using the same techniques?
Sound of crickets...
Re: (Score:2)
IANAP but there are things called optical computers and boson sampling is apparently a non-universally applicable type of Linear optical quantum computing.
https://en.wikipedia.org/wiki/... [wikipedia.org]
https://en.wikipedia.org/wiki/... [wikipedia.org]
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
Floating point arithmetic. This is solved by a special purpose coprocessor, which is now integrated into CPUs as an execution unit, but is omitted in microcontrollers where that functionality is not needed.
Re: (Score:2)
So far all the quantum supremacy claims have been "simulating" their own circuits. The equivalent of saying your camera and flash are better at computing photorealistic lighting and scenery than ray-tracing algorithms running on nVidia graphics cards. Technically true, but bullshit. Useless bullshit.
Re: (Score:3)
Yes, a camera is good at taking pictures, but it is not useful for any real-world computation tasks is it? The so-called quantum computer they built is good at simulating quantum phenomena, that's a true fact. But for actual useful computation, it is useless. My point is that they are basically running glorified experiments and claiming it is computation. An actual 100 qubit quantum computer should be able to do factorization. The camera can capture photo-real pictures instantly .. that's not useful for mak
So who are bigger liars (Score:2)
The world my never know.
How about doin some real world proof of supremacy (Score:2)
Instead of articles that kinda sensationalize and exaggerate these hypotheticals and all instead of claiming any company the winner in quantum supremacy I think they should prove their supremacy before attaining such a title. How about showing in public your quantum computer breaking something like RSA, Diffie-Hellman, ECC encryption algorithms, then post the work publicly. The first would show their temporary supremacy which would really put intense pressure on other companies in the field to do a bit of o