D-Wave Open Sources Its Quantum Computing Tool (gcn.com)

Long-time Slashdot reader haruchai writes: Canadian company D-Wave has released their qbsolv tool on GitHub to help bolster interest and familiarity with quantum computing. "qbsolv is a metaheuristic or partitioning solver that solves a potentially large QUBO problem by splitting it into pieces that are solved either on a D-Wave system or via a classical tabu solver," they write on GitHub.

This joins the QMASM macro assembler for D-Wave systems, a tool written in Python by Scott Pakin of Los Alamos National Labs. D-Wave president Bo Ewald says "D-Wave is driving the hardware forward but we need more smart people thinking about applications, and another set thinking about software tools."

Comments

Will this run on a Beowulf cluster?

[Anonymous Coward]

french toast, ladies

What??

[JustAnotherOldGuy]

"qbsolv is a metaheuristic or partitioning solver that solves a potentially large QUBO problem by splitting it into pieces that are solved either on a D-Wave system or via a classical tabu solver"

I know some of those words but all I can really tell is that it apparently does things to stuff, or does stuff to things.

Re:What??

[cfalcon]

> words

I'm not really sure those all quite qualify as words....

Re: What??

[MartyJG]

It's both words and not words at the same time.

Re:What??

[Anonymous Coward]

qbsolv: name of the program
partitioning: a family of methods for solving hard optimisation problems by splitting it up into subproblems
metaheuristic: a heuristic that generates heuristics
QUBO: quadratic unconstrained binary optimisation
D-Wave: the company that makes these machines
classical: non-quantum
tabu: A Tongan word, meaning "things that cannot be touched because they are sacred" (also spelled "taboo"). In this context, referring to tabu search, a common approach for solving hard optimisation problems on classical computers which involves marking certain potential solutions as "untouchable" if they've been seen recently, to avoid going over already-covered ground.

Now, the translation:

D-Wave is a Canadian company which makes quantum computers. It will either make (because it works) or break (because if Geordie Rose's hype implodes it will take the technology with it) quantum computing for a decade or two. The problem that D-Wave machines solve is a quirky one. It solves a specific optimisation problem, the details of which are unimportant for this discussion, but it's an optimisation (Ising spin minimisation) on problems that have a specific shape. This shape is the topology of the chip which D-Wave built. But all is not lost.

QUBO is an NP-hard optimisation problem that it is believed that certain classes of quantum computers (including D-Wave) can do especially well at, and it's also a problem that lots of real-world interesting NP-hard problems map onto in a reasonably clean way.

This program takes an arbitrary QUBO problem instance and tries to split it up into pieces that can either be run on a D-Wave machine or on a classical solver.

Re:What??

[Hognoxious]

"Ah, so it's kind of like facebook?"
--
The V.P. Of Marketing

Re: What??

[Anonymous Coward]

qwords

Be sure to drink your Ovaltine.

[0100010001010011]

• QUBO - Quadratic unconstrained binary optimization is a pattern matching technique, common in machine learning applications. QUBO is an NP hard problem.
• Tabu search take[s] a potential solution to a problem and check its immediate neighbors (that is, solutions that are similar except for one or two minor details) in the hope of finding an improved solution. Local search methods have a tendency to become stuck in suboptimal regions or on plateaus where many solutions are equally fit.

This won't help solving real world problems

[ffkom]

... any faster than can be done on conventional computers, today. All this hype around the D-Wave machines should not distract us from the fact that when competing for solving a given problem fastest against conventional computers, using the algorithms best suited for the respective hardware (and not making the conventional computer simulate a D-Wave like machine), the D-Wave machine loses every contest.
Open-sourcing some tools won't change that in any way.

Re:This won't help solving real world problems

[Richard_at_work]

Honest question then - why do people buy D-Wave systems then?

Re:This won't help solving real world problems

[ffkom]

Because D-Wave managed (via clever marketing) to scare a few organizations with deep pockets into thinking that if they didn't buy D-Wave systems, they could miss that once-in-a-lifetime opportunity to have a miraculous Quantum computer at hand while their competitors/enemies do not.

Re:This won't help solving real world problems

[Anonymous Coward]

And there's nobody as smart as ffkom at Google, Lockheed, or the national labs to realize they've been duped, so they keep putting more resources on the project and buying D-Wave's newer machines when they're available. :rolleyes:

Re:This won't help solving real world problems

[Pseudonym]

Here's a slightly more neutral answer:

Because D-Wave is a decade ahead of every other competing technology when it comes engineering and systems integration. D-Wave machines do actually work on real-world problems. They do not work as well as carefully-tuned classical approaches, but they do work.

It's possible that the D-Wave approach may be an evolutionary dead-end for quantum computing. Most people who understand the technology and are outside D-Wave (including Google and Lockheed) would probably put the odds at greater than 50% that D-Wave isn't the most promising approach. Nonetheless, we owe it to ourselves as an industry to test it because it's the only one that's here now.

Re:This won't help solving real world problems

[Anonymous Coward]

It's not necessarily about being faster, it can also be about finding new solutions to a problem. The new solutions may be more effective themselves. It's a "think outside the box" computing.

Re:This won't help solving real world problems

[ffkom]

The smart people at Google and Lockheed do realize that they can get funding for years of interesting research on all kinds of cool things by telling that "we are almost there with quantum computing", and money spent on under-performing D-Wave hardware is a rather small collateral damage in relation to the whole research funding riding the quantum computer hype, which may well result in useful findings, even if those are not quite the new super-computers that got them funded.

Fakable, and probably fake

[Anonymous Coward]

Take any process which can be simulated by a computer, and create optimized hardware to run that simulation. Encryption, 3D video, "quantum computing". Optimized hardware can outperform any general-purpose computer, even at lower clock speeds, especially if the operations don't map well to what is considered general-purpose. How much of D-Wave's positive reception is due to "ground-breaking quantum computing progress" vs. actual improvements in computation? I'm not saying that they haven't made real progress in running the algorithms quantum computers are claimed to do so well, I'm just saying there's no proof that any "quantum computing" is actually going on. While it isn't necessarily easy to do the faking, it's easy to cover it up when you don't let anybody look at the low-level hardware. While I expect the government to buy into this crap (they did, after all, fund psychic experiments for decades), I wonder what Google gets out of it.

Re:Fakable, and probably fake

[Pseudonym]

I'm not saying that they haven't made real progress in running the algorithms quantum computers are claimed to do so well, I'm just saying there's no proof that any "quantum computing" is actually going on.

Not true. Pretty much everyone who researches quantum computing is now convinced that the D-Wave machine does something that can reasonably called "quantum computation" because it matched a theoretical prediction. To a physicist, that's the best kind of evidence.

It's still not really that useful, of course.

Re:Fakable, and probably fake

[Anonymous Coward]

I do research in quantum computing and this is incorrect. The D-Wave machine appears to be using supercooled magnetic alignment, not quantum computing. The output of the predictive problem was computer slower than a classical computer and under simulation.

Re:Fakable, and probably fake

[Pseudonym]

Scott Aaronson disagrees with you, but OK.

marketroids

[John_3000]

"we need more smart people thinking about applications" = we have a solution looking for a problem

Re:marketroids

[silvermorph]

We have a hammer that no one can figure out how to swing.

Re: marketroids

[Anonymous Coward]

Exactly. Still a good tool, though.

"quantum" computing

[globaljustin]

D-wave is not quantum computing. It's regular, non-quantum computing that uses software to simulate what we think using non-locality in computing would be.

Humans actually controlling quantum non-locality would be arguably the biggest feat since harnessing the atom...it amazes me how this blatant bs continues to be called "quantum"...

Re: "quantum" computing

[Anonymous Coward]

Quantum annealing is a process that does use quantum processes on the chip, just isn't the usual quantum gate computer that get most of the news elsewhere (and have been actually build with very small numbers of gates... so you seem behind the news on that).

Re: "quantum" computing

[Anonymous Coward]

The D-Wave computer isn't BQP-complete, so you can't run, say, Shor's algorithm on it. It's much more like a quantum Enigma or a quantum calculating machine than a quantum computer in the ordinary, general purpose sense of the word.

Re: "quantum" computing

[globaljustin]

Quantum annealing

it's simulated programatically, it's not actual quantum annealing or any kind of actual non-locality (which like I said would be one of the biggest physics acheivements in human history)

Re: "quantum" computing

[Anonymous Coward]

actual non-locality (which like I said would be one of the biggest physics acheivements in human history)

You mean like quantum encryption schemes, long range entanglement tests, or even basic quantum gate computers (as already mentioned) that depend on entanglement to work? Nonlocality has already been in use for a while now. Whether or not those will end up in practical devices that are used outside of labs and startups is another story, but still seems like you are behind the times.

Re: "quantum" computing

[quax]

This is patently false. The machine has been confirmed to perform quantum annealing by independent researchers.

And actual non-locality is every day reality in experimental quantum physics, for instance for quantum encryption.

Re: "quantum" computing

[Anonymous Coward]

>This is patently false. The machine has been confirmed to perform quantum annealing by independent researchers.

The paper does not say this, and in fact in the conclusions admits that this is only some evidence but that classical computing could still be occurring.

Read your references before you post them.

Re: "quantum" computing

[quax]

The hedging in the paper is the typical verbiage that you get with any data driven study.

At any rate, it wouldn't be classical computing in the digital sense but reduce to a mostly analog annealer. The original claim was that the D-Wave machine was essentially a fake, and this has been discredited three ways till Sunday.

They set out to build a quantum annealer and it acts like one. What is unclear is how useful this process will actually be in practice. Quantum speed-ups are not at all guaranteed with this design. That's were the focus should be. Not some rehashed conspiracy theory that D-Wave is faking their hardware.

Re: "quantum" computing

[globaljustin]

simulated...it's all simulated

actual 'non-locality' would be faster-than-light travel and turn some core Einstein theories on their head...b?it would be transmitting information isntantaneously regardless of distance and that would be kind of a big deal

it's all simulated at some level...simulated or redefined in such a way to make it not actual non-locality

Re: "quantum" computing

[globaljustin]

What is unclear is how useful this process will actually be in practice. Quantum speed-ups are not at all guaranteed with this design. That's were the focus should be. Not some rehashed conspiracy theory that D-Wave is faking their hardware.

you're hitting it here...

the controversy is about what "fake" is...actual quantum non-locality would be faster than light information and that would turn Einstein's theories, as well as many other theories on their head

critics are right to say that D-wave is not actual quantum computing and it just hurts us all to let marketing rule in this arena

now, is D-wave "faking their hardware"? in essence, no, it's real software on a machine that exists in reality...but it's not "quantum" like they are claiming

everyone wants to race to being able to say "quantum" and predictably people are cheating on how they define it to claim to be the first...it's not a new tactic, but we should still be critical of it!

Re: "quantum" computing

[quax]

The ERP paradox that you are alluding two was solved by Niels Bohr just two days after Einstein presented it to him.

Quantum Information is fundamentally different than classical Information. A pure quantum information channel that only establishes entanglement does not transport classical information, it just established a quantum correlation. But this correlation can only be confirmed after corresponding classical information over the measured state on one end of the channel reaches the other end.

That classical signal of course obeys relativity and cannot propagate faster than light. Hence causality is preserved.

The common statement, (that many physicist who should know better make as well), that entanglement establishes an instantaneous signalling is grossly wrong. For spacelike separated events, linked via a quantum channel, instantaneous is not even defined. It is completely meaningless as the temporal order is relative to an observer's inertial frame. So I guess I shouldn't say it's grossly wrong, but stick to Pauli's formulation. It's not even wrong, it's worse, it carries no scientific meaning whatsoever.

(BTW I am physicist and dabble in quantum computing).

Re: "quantum" computing

[Anonymous Coward]

actual 'non-locality' would be faster-than-light travel and turn some core Einstein theories on their head

redefined in such a way to make it not actual non-locality

You're the one doing the redefining here. Non-locality, as originally defined by the EPR paper, has been demonstrated to the point that only alternative is superdeterminism, which is impossible to rule out completely. If you want to redefine it to mean something else, either qualify it, or expect to look like an idiot who doesn't know what the word means.

that would be kind of a big deal

Yeah, it was, which is why it contributed to the 2012 Nobel prize and other prizes like Wolf prizes, etc. But after a decade, it became less big news every time someone increased the distance another 10 or more kilometers than the last guy.

Re: "quantum" computing

[globaljustin]

entanglement establishes an instantaneous signalling is grossly wrong

look at the tachyonic anti-telephone: https://en.wikipedia.org/wiki/...

using non-local quantum entaglement to instantaneously transmit information indeed would be faster-than-light

humans have not achieved non-local entanglement, nor have they used it for computing, and unless you redefine "quantum" then it's not possible for these to be "quantum computers"

yes, you're right that we could only verify the signal transmission at light speed, but if it is instantaneous then it's instantaneous...if the phenomenon happens and we verify it then the ability is confirmed...time is more than just what any one observer sees and we can determine if something was instantaneous vs faster-than-light...why couldn't we?

I'm just a telecommunications scientist so this is definitely out of my expertise in some ways, but I think I know enough to understand that true non-locality has not been achieved and that for a "quantum" computer to truly be quantum, it would have to use actual instantaneous non-locality to process

thanks for this discussion it's been helpful

Re: "quantum" computing

[quax]

"Using non-local quantum entanglement to instantaneously transmit information indeed would be faster-than-light."

Yes, and entanglement can't do that as I tried to explain in my earlier post.

You do not transport information but quantum information with entanglement. They are as far removed from each other as Schroedinger's cat from any pet you've ever owned.

BTW tachyon are hypothetical faster than light particles. Of course you could signal into the past if such a resource existed and could be technologically controlled.

And once again, nature has no concept of "instantaneous" it's a human approximation to a constant frame without relativistic speeds. Whenever a physicist uses this word in connection with entanglement its an unscientific shortcut that has no real physical meaning.

I'd highly recommend you to draw up some Minkowski diagrams, until this clicks for you.

Understanding the nature of quantum information is a bit more tricky, because it all comes down to how quantum correlations will look like perfect random noise until you get the information from a corresponding measurement on entangled systems. The way this is experimentally tested is via Bell's inequality violations.

Bell was much later than Einstein. Bell essentially codified the latter's reservations and wanted to prove Einstein right, but nature did not cooperate.

BTW the only quantum spookiness that QM allows are Quantum Erasers.

And please note, that all these oddities are not hypothetical but demonstrated in experiments. It's these weird experimental results that have brow beaten physicists into accepting entanglement as a fundamentally non-local phenomenon, against the marked resistance of some of the best and brightest minds like Einstein and Bell.

Re: "quantum" computing

[globaljustin]

I will definitely have a look, thank you very much.

There is something here that I think is relevant:

Bell essentially codified the latter's reservations and wanted to prove Einstein right, but nature did not cooperate.....It's these weird experimental results that have brow beaten physicists accepting entanglement as a fundamentally non-local phenomenon, against the marked resistance of some of the best and brightest minds like Einstein and Bell.

Now, I don't expect a response necessarily because I'm just spitballing here (and you've been very kind thusfar attempting to explain your point to me)....but I seem to recall from my previous look into "spooky action at a distance" that the idea that Einstein and Bell had been "proven wrong" was generally accepted, but only if we make some assumptions about how the universe works that, while generally accepted as true, are dependent on a definition of some concepts that are abstract and not scientifically provable.

I know...read your links...and I will (thanks again)...but I just can't shake the idea that I've taken a deep dive into this before and found that there are still some fundamental assumptions that are a sort of philosphical scaffolding for proving Bell/Einstein "wrong" on action at a distance.

How well does this thing perfom...?

[jgfenix]

... compared to a normal computer or a GPU? And what about the energy consumption? Is it useful more than the coolness of the quantum thing?

Re:How well does this thing perfom...?

[Anonymous Coward]

It performs at about the same performance to ~600 times more slowly, depending on the problem instance. See Scott Aaronson's d-wave blog entries here, or discussion from a previous Slashdot thread.

Re:How well does this thing perfom...?

[Anonymous Coward]

Ignore the ACs. The initial versions were slower, because it is new tech. As qbits are added the scale of performance is not linear but more like exponential, since well, it is using qbits for computation.

This was and is known as a given by the manufacturer and early adopters.

we have a solution looking for a problem

[Dusthead Jr.]

People seem to say that as if it were a bad thing. Personally I'd rather have solutions to problems I've yet to have. Inversely, I wouldn't mind having all of my future problems solved using only existing solutions. A win-win in my book.

Re: we have a solution looking for a problem

[John_3000]

It's a bad thing in that it's bad business --- a bad investment because no one needs to buy the product. Fine as a hobby though, or as research if you have a grant.

Hammer

[hackwrench]

We have a hammer that does interesting things when swung. We want to study this hammer to see if what we learn can't be used to make a hammer that is actually better than the best hammers out there.

CMAES better

[Anonymous Coward]

I'm disappointed to find it splits up the problem, you rarely find an optimal solution by trying each variable as though they're independent.
And I'm most disappointed that it's unconstrained. Which makes it totally useless for every problem I've ever used an optimizer.

CMAES (available in various libraries, e.g. Apache Math) is probably a better choice for "can't solve it any other way" optimizations. That tries N points randomly distributed across each variable depending on each variables deviation. With N based on solid probability, ( but still may not find the perfect solution).

You'd think that they wouldn't have to split it up, with 'quantum fakentanglement' linking the variables at faster than the speed of light... but then what do I know. I only know statistics for data mining, common faults in datamining (like prefiltering for linking variables on apparently uncorrelated data), montecarlo tests and so on.... just math and science not quantum magic.