9:00 PM
@vzn I don't see how is this computation supposed to prove QC supremacy to classical computers. They can claim that only when they perform a useful computation faster than a classical computer.
@AccidentalFourierTransform But you said yourself that you believe premarital sex is a sin... ;)

@Mithrandir24601 "beat" is roughly translated as "quantum supremacy" a relatively new term that was used in the article cited... do understand many of the complexities hence my skepticism. there are many different reasonable metrics... & dont think we are close to any of them "yet"...

@Mostafa it sure is
have you never read the bible?

> actually lately am nervous that theyre (mostly Martinis!) overpromising at this pt :|
I don't suppose you wish to explain why?

@Mostafa exactly, thats roughly my pt & where think the "overpromising" is starting to show up. but martinis has an expensive lab & has to be a salesman sometimes... :|

@ACuriousMind I would certainly not call my little group "Google's physics department".
@Mostafa The name "quantum supremacy" is generally disliked by everyone in the group.

9:04 PM
Oh?

We didn't invent it. Most of us really, really wish we had a better name for what we're doing.

@DanielSank (uh oh lol) turning that around... what do you think you will be doing with 49 qbits at the end of the year? in terms of computations?

@vzn Nice try. Why don't you either explain why you think we're overpromising or take it back?

@DanielSank I'd have bet there are universities with physics departments of a similar small size.

@DanielSank do you have an alternative? we can try to use it... noticed that "supremacy" is not liked by everyone in the field, saw a blog on that lately etc

9:05 PM
@ACuriousMind Certainly true, but Google is rather not small...
@vzn Why are you dodging the fact that you accused my colleagues of overpromising?

@DanielSank ok its not clear what the real msg is in meda vs martinis, but he seems to be connecting "demonstration of quantum supremacy" with "49 qbits at end of year"...

I do believe that is bordering on violation of "be nice" if you don't explain your comment.
You are wandering into baseless accusation territory.

@DanielSank sigh dont want to argue try to avoid the adversarial stuff :(

@vzn Are you aware that "quantum supremacy" has a very specific meaning?
@vzn Then either 1) Don't accuse people of stuff or 2) Consider this a discussion and not an argument.

@vzn Well, don't start to doubt other people's statements if you're not prepared to argue.

9:07 PM
@DanielSank it means something, what does it mean to you? it was the headline in an article that presumably interviewed martinis etc...

@vzn Nice try. Why don't you answer my question first?

@AccidentalFourierTransform You're not married and therefore shouldn't have sex. So why don't you wear pants?!

because being naked is certainly more comfortable
fuck pants

O_o~"

9:10 PM
@DanielSank hope you know me well enough to understand that am rooting for your team and have been since the very beginning, one of the worlds biggest qm computing/ google enthusiasts, have blogged on it extensively etc...

@DanielSank Sooo...what would you call it? Google's tiny physics investigative task force?

@vzn You are, again, avoiding the main thrust of the discussion.
I am not questioning your rootingness.
I asked you if you are aware that "quantum supremacy" has a specific meaning.

@DanielSank youre avoiding some stuff too

No, I'm not.

@DanielSank I'm actually really curious about what it means by 'execute operations on the array that will cause it to evolve chaotically and produce what looks like a random output', but I'm expecting you to not tell me yet :/ although I wouldn't complain if you did :P

9:11 PM
I'm trying to keep the conversation on one thing at a time.

@DanielSank like said, its a relatively new term etc... no rigorous/ strict meaning yet afaik, multiple reasonable ways to define it etc

That's true.
What we mean by that term is this: you have a programmable system where you cannot predict the outputs with a classical computer.
That does not at all say that the quantum system is "better" or "more useful" than the classical one.

@DanielSank ok, thats a bit different than a popsci use of the term or many other usages of it afaik.
did you see this?
> By the end of 2017, Google hopes to make a 49-qubit chip that will prove quantum computers can beat classical machines

@vzn Well, "beat" is disappointingly vague there.
I would not use popsci journalism as a reference point for the meaning of scientific terms.

> By the end of this year, the team aims to increase the number of superconducting qubits it builds on integrated circuits to create a 7-by-7 array. With this quantum IC, the Google researchers aim to perform operations at the edge of what’s possible with even the best supercomputers, and so demonstrate “quantum supremacy.”

9:15 PM
xDDD
If you were my roommate, I'm sure we'd both fail all the courses together xD

@vzn That sounds perfectly reasonable and does not say the quantum chip is more useful than the classical one.
@vzn No. I don't usually read the popsci magazines.
I probably should.

experiencing cognitive dissonance

do you know what the "best supercomputers" can compute per time? & think the 49 qbits are going to come close?

@vzn This was actually first reported about 2 months ago, so isn't really anything new...

9:20 PM
@Mithrandir24601 right, its still highly questionable 2 mo later :P

@vzn The whole point of the 49 qubits is that you can't simulate them on a supercomputer because of the memory and interconnect requirements.

@DanielSank understand your definition of quantum supremacy. its different than the article. its a disconnect.

It's not different than the article. The article is vague.

@DanielSank this is where it seems to me even scientists get a bit mixed up, dating all the way to feynman. simulating qubits is far different than general computation...

@vzn Who said anything about general computation?

9:22 PM
@DanielSank thats the pt, a supercomputer performs general computation

@vzn Yes.
and?
Nobody is saying the quantum chip is better or more useful than the classical computer.

@DanielSank lol the article is

I doubt that.
And even if it did, that would be one journalist making a sloppy mistake.
That's rather different than our group overpromising stuff.

anyway think this exchange demonstrates some of the subtleties involved & dont expect to resolve anything here. (& dont think its attributable to "merely" sloppy journalism)

@DanielSank I have not studied much about quantum computing, but always have one question whenever see the news about it: why do you think building a 7x7 array of qubits (using an extensive and expensive infrastructure) means even a small step toward realization (and even commercialization) of quantum computers, while the number of the required qubits to perform a useful task is at least hundreds of millions?

9:25 PM
@vzn Is it questionable that they're 'aiming' to get a 7x7 qubit chip? Or is it questionable that said 7x7 chip will demonstrate 'supremacy'? They're obviously 'aiming' to get one - whether or not they will is a different question (fab might be slow, might have some random minor design problem that slows the production process down slightly etc. are possible, but that doesn't mean they're not 'aiming' for it)

@vzn Have you noticed that you very often get into discussions where someone else here is asking you to be more careful with your language?
It's a pattern.
I advise you to take note of it.
@Mostafa The 7x7 grid is a big deal for a few reasons:

@Mithrandir24601 7x7 is achievable. general computations on level of supercomputers are not "near"

1) Making 2D grids is a huge engineering challenge. 2D grids are needed for the million-qubit system, so it's important to show we can do it.
2) With ~50 qubits, you can do stuff that you can't simulate on a classical computer, so it's at least a suggestion that the quantum chip is doing something special/new.
3) With that many qubits, we it might be possible to actually beat a classical computer on a useful problem, e.g. chemistry stuff.

@DanielSank asked you to come up with a better term than "quantum supremacy" which you say you didnt like (already understood it was problematic within field) & you didnt respond.

@vzn That is correct. I fail to see how that is a response to me pointing out your repeated tendency to make unfounded claims.
But anyway, I would say something like "quantum unsimulateability".
But I'm not a pop-sci writer.

9:29 PM
@DanielSank think you are too adversarial sometimes. am going elsewhere now. :(

@vzn That's what @DanielSank was just talking about... In this case, "at the edge of what's possible" isn't the same thing as "this can do everything better than a supercomputer" but rather "there exists at least one thing that this can do in a time at worst comparable to that of a supercomputer" or something to that effect

Apparently, to ask another user to provide support for a claim that my research group is overpromising results makes me adversarial.
@ACuriousMind was vzn's statement "offensive"?
It was something of a personal attack, albeit a rather tempered one.

@DanielSank Which one? (neither of them was offensive in my view, they're just mostly unsupported)

I see.

And I don't think you were adversarial, at least not more than these claims warranted.

9:42 PM
@Mithrandir24601 I think the problem is that some believe if we could ever build large enough quantum computers (to perform useful computations faster than currently available supercomputers) that wouldn't happen anytime in the foreseeable future.

@ACuriousMind Certainly not.
@Mostafa It's really not obvious yet.
A universal gate-model quantum computer is some time off. However, I would not call it beyond forseeable future.
The progress in our field is rather encouraging.
We're a bunch of physicists and we're building this ridiculously complicated microwave frequency analog computer and it works.
Go tell the world in 1990 that you can make an electrical circuit quantum mechanical and enjoy the skeptical laughs.
Nobody thought that even the working principle was possible!
Yet here we are, doing logic gates with high enough accuracy to conceivably build a universal computer.

@Mostafa For a start, what do you mean by useful? There could be other quantum analogue things that could potentially demonstrate this 'quantum supremacy' thing before a properly useful large (fully connected) quantum computer comes along

Heck, for all we know you can invent new catalysts without a fault tolerant quantum computer. We just don't know!
IMHO everyone is throwing around predictions and skepticism way too early.
We have to build something, give it to the public, and let smart people screw around with it to see if it's useful.
Just like every other piece of technology ever.

Theoretically, it appears that it should be useful though...
(assuming you can build a properly large, fault tolerant etc. etc. quantum computer)

@Mithrandir24601 Even then, what do we know we can do with it?
1) Grover
2) Shor
3) ???

9:51 PM
@DanielSank Boson sampling, obviously

@DanielSank Improve the speed of e.g. routing problems: journals.aps.org/pra/abstract/10.1103/PhysRevA.95.032323

@Mithrandir24601 Most of those problems are not known to not have faster classical algorithms.
We think they don't, but we're not sure.
It's exciting!

@EmilioPisanty The entire point of boson sampling is to demonstrate this 'quantum supremacy', so I don't think that counts :P

Then after that, endlessly repeat "because nature is quantum, dammit" on any and all conference slides.

@Mithrandir24601 By useful I meant running traditional tasks like factorization of large numbers, or simulation of large (quantum mechanical) many body systems; works that are virtually impossible for a classical computer.

9:53 PM
@Mithrandir24601 fyi a very revealing/ widely applicable concept behind technological advancement en.wikipedia.org/wiki/Hype_cycle

@EmilioPisanty Also, fully 'solving' boson sampling is actually #P

@Mithrandir24601 hmmmmm. That kinda depends on what you take 'solve' to mean, no?

@Mostafa Does simulating the behaviour of n qubits using an n qubit computer count then? :P

I mean, the original formulation of the problem was carefully chosen so that it would not be #P hard

@vzn Yep - I've heard of this quite a bit. The question is: where are quantum computers on the graph? Well, we're at the 'startup companies' bit, but not yet at the 'first-gen' bit, so... We don't have a clue
@EmilioPisanty In this case, it's (I think) finding the probability distribution of measurement outcomes

9:58 PM
@Mithrandir24601 glad youve heard it because its unfamiliar to many. it shows up esp in technology but is applicable to science also. lived thru dotcom frenzy myself. in a sense a technology can be at more than 1 pt of the graph/ curve at a time...

What the hell is #P?

Yes if you're *really* interested in the behavior of an n-qubit quantum system.
I specifically had material science simulations in mind (dealing with at least hundreds of electrons)

@Mithrandir24601 yeah, precisely, which is why boson sampling is restricted to just sampling from the distribution
@ACuriousMind something something count solutions something something
big hard ugly beast

@DanielSank Simulating quantum systems

The introduction to the Wikipedia article is pretty much indecipherable for me.

9:59 PM
@ACuriousMind complexity class that contains NP

@ACuriousMind it's essentially taking an NP problem that asks "is there a foo that does bar?" and turning it into "how many foos do bar?"
(take with a grain of salt, my grasp on complexity theory is pretty shaky)
@ACuriousMind is it really?
In computational complexity theory, the complexity class #P (pronounced "number P" or, sometimes "sharp P" or "hash P") is the set of the counting problems associated with the decision problems in the set NP. More formally, #P is the class of function problems of the form "compute ƒ(x)", where ƒ is the number of accepting paths of a nondeterministic Turing machine running in polynomial time. Unlike most well-known complexity classes, it is not a class of decision problems but a class of function problems. == Relation to decision problems == An NP decision problem is often of the form "Are there...
I think the examples are pretty ilustrative

@EmilioPisanty Hmmm, that one is better

@ACuriousMind And more importantly as far as boson sampling goes, calculating matrix permanents is #P hard
The #P-completeness of 01-permanent, sometimes known as Valiant's theorem, is a mathematical proof about the permanent of matrices, considered a seminal result in computational complexity theory. In a 1979 scholarly paper, Leslie Valiant proved that the computational problem of computing the permanent of a matrix is #P-hard, even if the matrix is restricted to have entries that are all 0 or 1. In this restricted case, computing the permanent is even #P-complete, because it corresponds to the #P problem of counting the number of permutation matrices one can get by changing ones into zeroes. Valiant...
↑ actually has a wikipedia page of its own

@ACuriousMind it helps to start with understanding NP...
#P seems to have been mostly rather obscure (maybe not even analyzed/ considered much?) until todas thm en.wikipedia.org/wiki/Toda%27s_theorem

10:14 PM
@EmilioPisanty I see.

@ACuriousMind also related:
Immanant redirects here; it should not be confused with the philosophical immanent. In mathematics, the immanant of a matrix was defined by Dudley E. Littlewood and Archibald Read Richardson as a generalisation of the concepts of determinant and permanent. Let λ = ( λ 1 , λ 2 , … ) {\displaystyle \lambda =(\lambda _{1},\lambda _{2},\ldots )} be a partition of ...
just saw that
and yes, that'd be an interesting thing to do to a matrix, now that you mention it
(to wit, $${\rm Imm}_\lambda(A)=\sum_{\sigma\in S_n}\chi_\lambda(\sigma)a_{1\sigma(1)}a_{2\sigma(2)}\cdots a_{n\sigma(n)}$$ for $\chi$ a character of $S_n$)

@EmilioPisanty Would it? Is that even basis-invariant?

10:51 PM
@ACuriousMind looks like
it's almost certainly a group homomorphism
same proof as the determinant
all you needed was that $\chi=\mathrm{sign}$ be a homomorphism
which is precisely what you're imposing
and if it breaks matrix products into complex multiplication, basis changes go away
so yeah