Google claims to have proved its supremacy with new quantum computer • Hacker News Board

Re: Google claims to have proved its supremacy with new quantum computer

#101

Post by abdullahkhalids » Mon, Jul 03, 2023, 6:58 PM UTC

alpineidyll3 wrote: post #94

abdullahkhalids wrote: post #87

Earlier quoted context omitted.

Not for photonic quantum computing. Only detectors require cooling, and it is possible to build adequately sized quantum computers with constant number of detectors using loop based architectures. Even more realistic architectures are very very cost effective on the number of components https://quantumfrontiers.com/2023/06/21/what-is-the-logical-...

Yes, but all existing photonic platforms use post-selection which is even more clearly exponentially lossy. Although this could be solved with a deterministic single photo source if one can be found. Photonic quantum computing is an especially funny post-transistor paradigm because classical photonic computing is also quite attractive.

If you read the linked article, it explains how to avoid exponential loss on post-selection.

> Imagine you can toss coins, and you need to generate 20 coins showing Heads. If you repeatedly toss all 20 coins simultaneously until they all come up heads you’d typically have to do so millions of times before you succeed. This is even more true if each coin also has a 20% chance of rolling off the table (akin to photon loss). But if you can toss 20 coins, set aside (switch out!) the ones that came up heads and re-toss the others, then after only a small number of steps you will have 20 coins all showing heads. This large gap is fundamentally why the first whammy is not relevant: To generate a large photonic entangled state we begin by probabilistically attempting to generate a bunch of small ones. We then select out the success (multiplexing) and combine successes to (again, probabilistically) generate a slightly larger entangled state. We repeat a few steps of this. This possibility has been appreciated for more than twenty years, but hasn’t been done at scale yet because nobody has had a good enough optical switch until now.

Re: Google claims to have proved its supremacy with new quantum computer

#102

Post by s1dev » Mon, Jul 03, 2023, 6:59 PM UTC

alpineidyll3 wrote: post #75

AbrahamParangi wrote: post #58

Earlier quoted context omitted.

There's a fundamental question as to whether or not it's exponentially difficult to add additional qubits. If each marginal qubit is 5% more difficult to add as the previous one the task would be essentially impossible - and even if it wasn't impossible it would turn out to basically be cheating (in that you would be doing exponential work via either the quantum computing route or the conventional route).

Note: that there is good physical reasons why the cost of QC may grow exponentially with qbits. Refrigeration is exponentially inefficient as T=>0 and the gap of a quantum system which sets the temperature you must cool to shrinks as you couple new degrees of freedom. This dynamic has been the basic reason for the sub exponential progress in the area (despite exponential expenditure)

This is not quite true. You only need to keep the qubits at a fixed temperature as you scale the system, so the resources required to add additional qubits grow only polynomially with the system size. Once you have many qubits with a sufficiently low (but constant) error rate, you can do quantum error correction which also only has polynomial overhead.

Re: Google claims to have proved its supremacy with new quantum computer

#103

Post by dotnet00 » Mon, Jul 03, 2023, 6:59 PM UTC

taeric wrote: post #43

knorker wrote: post #26

Earlier quoted context omitted.

Possibly. This is way beyond my knowledge. But I'm not aware of many other "search this finite part of the number line for this property", where "this finite part" is still too big for classical computers. It almost sounds like quantum computers are tailor made for the types of problems we've been building cryptosystems on. But maybe this is not all quantum computers will be able to do. I couldn't even explain exactl…

"Search this finite part of a number line" is probably fairly accurate in describing many Mixed Integer Programming (MIP) techniques. Such that, if you do get a breakthrough at that, you may see larger advances in optimization.

Exactly, ML is a potentially very valuable task for more mature quantum computers due to the ability to perform much more powerful searches than what we have to accept on classical computers.

Re: Google claims to have proved its supremacy with new quantum computer

#104

Post by dataflow » Mon, Jul 03, 2023, 7:00 PM UTC

carabiner wrote: post #91

dataflow wrote: post #79

Earlier quoted context omitted.

> The best way to get a fast intuition for why the simulation is superior is to take an entry-level CAD course with a focus on material design. I'm sorry but you're completely missing the point of my question. The question was not "what can simulations do that experiments can't". My background in simulation is not zero, and I've never had that question. The question as something else entirely. > You can’t measure a n…

Are you just asking if these are unrealistic, contrived performance benchmarks? Those have always existed and they're fine: https://news.ycombinator.com/item?id=20231084 Some are more useful than others. There's no strict criteria just as there is no perfect way fully characterize CPU performance.

"Unrealistic" or "contrived" don't really get to the heart of the matter, at least not as I understand them. The stick example can be extremely useful and realistic - it doesn't really make any difference to the question.

Perhaps another way to phrase it (though I'm not 100% sure this is equivalent) is: how do they know whether whatever they're accomplishing is quantum computation, as opposed to something else (like analog computation) that happens to be concerned with quantum physics.

Re: Google claims to have proved its supremacy with new quantum computer

#105

Post by pastelsky » Mon, Jul 03, 2023, 7:00 PM UTC

If I understand correctly, the limited amount of "expressiveness" a quantum computer has restricts its ability to solve useful problems.

Couldn't you simply express useful problems as a function of problems that the quantum computer can already solve? Doing so might be extremely in-efficient, but give that there's so much performance leeway, it still might end up being similar to super computers of today?

Re: Google claims to have proved its supremacy with new quantum computer

#106

Post by pyrale » Mon, Jul 03, 2023, 7:06 PM UTC

knorker wrote: post #9

taeric wrote: post #3

"This is a very nice demonstration of quantum advantage. While a great achievement academically, the algorithm used does not really have real world practical applications, though." Not having real world applications is not necessarily damning, of course. Curious to know what implications this has for general algorithms. Reading this, it almost makes it sound like there will be some algorithms that quantum is better a…

My understanding is that quantum computers only have two real use cases, as of today: 1. Breaking crypto. 2. Simulating other quantum systems. For (1) it's basically all downsides. For (2) unless you're a particle phycisist you'll never need quantum computers. But that's now. Maybe there will be a killer app for it some day, changing everything. Or indeed, we could get it indirectly. Maybe simulating quantum systems…

Basically, quantum computers can solve problems of the BQP class in a time that is a polynomial function of input size. These same problems, on a traditional computer, can take exponential time.

It is hard to tell with certainty what can be done with quantum computers, but having powerful quantum computers will certainly open new fields of research, like trying to figure out if quantum algorithms for NP-class problems have better complexity than classical algorithms. Simply finding subexponential algorithms for problems that would have required exponential time on a traditional computer, may prove valuable.

Basically, outside of the immediately available problems, this opens up an entirely new complexity class from which to attempt to crack other classes of complexity. Trying to break NP-problems with turing machines, if not successful in an academic sense, has proved immensely valuable and profitable. By getting another architecture, we get a shot at similar breakthroughs.

Re: Google claims to have proved its supremacy with new quantum computer

#107

Post by junon » Mon, Jul 03, 2023, 7:09 PM UTC

Gravyness wrote: post #83

bob1029 wrote: post #6

Earlier quoted context omitted.

> The theoretical basis for these experiments depends on sampling the output distributions of random quantum circuits; unfortunately, understanding how this theoretical basis can be used to define quantum supremacy is an extremely difficult task. Anyone attempting to understand how this sampling task relates to quantum supremacy must study concepts from random matrix theory, mathematical analysis, quantum chaos, comp…

> This might be a case of a rabbit you do not want to chase. I'm sure the researchers will all collectively realize this, if it were to be the case, and disregard their 12 year academic journey with their great salaries in favor of research into more important topics like world hunger, renewable energy, etc. I'm so sure that the percentage of sureness is an imaginary, quantum-entangled value between -7 and 13 billion…

"academic journey" and "great salaries" generally don't go together.

Re: Google claims to have proved its supremacy with new quantum computer

#108

Post by reb » Mon, Jul 03, 2023, 7:10 PM UTC

BasedAnon wrote: post #62

zdragnar wrote: post #27

Earlier quoted context omitted.

The difference between you breaking a stick and the computer modeling it is that you've measured nothing. You don't know, with any precision, the amount of force you used, the rate the stick broke at, how much mass remains in the two pieces and how much was lost to splintering, etc. In other words, assuming the computer model has sufficiently accurate data as an input, it can produce significantly more refined output…

Now if we had a way of measuring it, it would be interesting to ask what sort of model of computation we could derive from the breaking of sticks

I cannot wait to tell you about the cutting-edge technology my team is developing — it's called the twigchain and it's going to change everything.

Re: Google claims to have proved its supremacy with new quantum computer

#109

Post by robot_no_421 » Mon, Jul 03, 2023, 7:14 PM UTC

ori_b wrote: post #98

robot_no_421 wrote: post #88

Earlier quoted context omitted.

You're not modeling or predicting anything though. That's like saying "what if I built a bridge that failed on the first day? A computer would need several days to calculate all the forces that led to the failure, but my bridge failed just fine without any computer help". Well... yes. But try building a bridge that doesn't break. Or to keep with your scenario, try to predict exactly where and how your stick will brea…

Does Google's implementation of quantum computing help with this sort of scenario, or is it a really fancy way of breaking the bridge?

Yes, and no. Don't ask me rhetorical questions if you don't want a rhetorical answer.

Re: Google claims to have proved its supremacy with new quantum computer

#110

Post by kelnos » Mon, Jul 03, 2023, 7:16 PM UTC

dataflow wrote: post #79

JumpCrisscross wrote: post #70

Earlier quoted context omitted.

> I can literally measure those with a ruler and a scale You can’t measure a number of internal stress-strain conditions during the moment of failure. You can’t repeat the experiment with the same stick. The best way to get a fast intuition for why the simulation is superior is to take an entry-level CAD course with a focus on material design.

> The best way to get a fast intuition for why the simulation is superior is to take an entry-level CAD course with a focus on material design. I'm sorry but you're completely missing the point of my question. The question was not "what can simulations do that experiments can't". My background in simulation is not zero, and I've never had that question. The question as something else entirely. > You can’t measure a n…

> What if all I care about is something easily measurable, like the length of each remaining piece?

If you're not just talking about the two large pieces, but also the lengths of the small splintered pieces, I suspect the computer is going to be a lot faster at figuring that out than you will be manually measuring them.

Your responses in this thread seem to be a bit disingenuous: first you ask why a computer simulation could be better than doing it manually, but then when people tell you the things the computer can do faster/better than you, you say "but what if I don't care about that?" Well, duh, if you don't care about anything but the most trivial things, the computer probably isn't going to be of any benefit to you.