Maybe you should spend some more time on the D-Wave site and actually read it?
On the page you link to under the "What you will learn" section:
How to use the D-Wave OneTM System as a co-processor to a conventional computer in a scalable way.
The D-Wave chip is a special purpose solver, it relies on a classical processor for loading and pre-processing, that is where this python code gets executed.
Anyhow, if you want to learn what's on the chip check out this section.
Programming the D-Wave chip is nothing but initialising the spin states of the qubits, it has nothing to do with classical transistor logic. And no, it is not alien technology, although apparently pretty foreign to you.
Yet, you write like somebody who already made up his mind, and I doubt you're willing to learn anything that contradicts your preconceived notion.
The contraptions Rossi builds are cheap and look like a plumber put them together. On the other hand D-Wave has chips samples on display that are produced by a special purpose foundry that can produce Niobium SC circuitry. That took some serious investments.
Rossi supposedly sold his house to finance his venture, D-Wave is backed by the likes of Steve Jurvetson and Jeff Bezos. Rossi has phantom customers who don't talk to the public, D-Wave has Google, NASA and Lockheed Martin.
Frankly, after they started publishing in Nature it's ludicrous to hang on to the idea that this is just an elaborate fraud. Then again conspiracy theories are a dim dozen on the Internet, and you are free to believe whatever you want.
Sorry to blow your bubble, but you can actually look at the chip that's inside D-Wave's boxes. It doesn't have transistors, it has Josephson junctions. Tell me how you can get any calculations out of those unless you do physical annealing with them.
But to hang on to this stance after the amount of scrutiny that the D-Wave machine received is about as rational a climate change denial.
It's one thing to argue that they have not proven a quantum speed-up, but they clearly build an quantum annealing device that you can use to perform calculations.
I was referring to the paper by Mathias Troyer et. al. that is yet to be published not the effort that Alex Selby writes about (thanks for the link).
Will have to read the latter in more detail to get a good grasp of how much effort is required to beat the benchmark with Alex's approach. Two caveats: On first glance I am not sure if he has the same training data (he mentions he communicated with Cathy McGeoch) - if he does it'll be interesting to see how stable his generic approach is when the problem domain is slightly altered.
The problem is mostly on the engineering side, it'll be hard to build up industrial capabilities of the scope needed for a useful universal gate based QC and conceal if from the rest of the world. It's not like coming up with a code cracking algorithm, the latter could be kept classified quite easily.
That'll be a lot of effort for a fraud, especially since you then have to fake all the qubit specific data that goes into the publications. And the chip samples they have on display very much looks like Josephson junctions circuits and nothing like regular chips. (And the integration density they have for this process could not at all deliver reasonable classical performance).
Ahem... the central claim that their chip is a true quantum chip leveraging qubit entanglement has been demonstrated i.e. this paper.
Matthias Troyer, one of the co-authors, expressed in an email to me, that he was surprised to see this evidence, but that the chip seems indeed to perform some sort of quantum annealing.
You may think they are misguided, but their tech is for real. Even Scott Aaronson doesn't deny that.
There is no classical computer hidden inside, but there is still reasonable doubt as to exactly how quantum the device is, and if it will ever deliver clear cut quantum speed-up.
No, the D-Wave machine can only solve the Ising equation. Universal adiabatic quantum computers have been shown to be able to emulate gate model quantum algorithm, but for the more restricted current D-Wave architecture a mapping is (probably) not possible. Nevertheless the class of problems they can solve is still pretty large, and is applicable to useful optimization use cases and learning algorithms.
The benchmark did indeed not demonstrate a quantum speed-up, but it in fairness to D-Wave this was a test designed based on the customers requirements i.e. for them acing this benchmark was good enough to justify investing in this technology.
My understanding is that the algorithm that was comparatively fast on a classical computer was hand optimized by a graduate student, it was not a generic annealing algorithm solver.
But the paper on this effort of 'beating' D-Wave on a classical machine is yet to be published, so this is all from blog hearsay.
If it was only shoddy experiments that'll be one thing. But some theoretical physics has gotten so used to divorcing itself from experimental verification it's turning into fairytale science.
Reputable journals are only marginally better. Just witnessed a back and forth where some research was attacked by a prominent scientist. The assumptions the latter made weren't quite on target, so the attacked researchers submitted a paper pointing this out. This passed anonymous peer review but then the paper solicited the opinion of this star scientists. He dismisses the paper with the most bizarre arguments that give the impression that he didn't even read it. Then the prestigious journal turns around and endorses this position and rejects the paper.
Absolute astounding and sobering.
Disclaimer: I am not a party or author in this but covered the controversy on my blog, and hence have been shown the rejection email.
Flat earthers, creationists and climate change deniers are immensely annoying but this is still a terrible decision. Trolls will never go away, but they are only kept in check by challenging them.
On the other hand science can benefit from intelligent and and well informed commentary and like any other human activity benefits from critical, constructive media coverage and public debate.
Slashdot Mirror
Maybe you should post some of your concerns to this reddit group that has people who actually work with D-Wave boxes.
Maybe you should spend some more time on the D-Wave site and actually read it?
On the page you link to under the "What you will learn" section:
How to use the D-Wave OneTM System as a co-processor to a conventional computer in a scalable way.
The D-Wave chip is a special purpose solver, it relies on a classical processor for loading and pre-processing, that is where this python code gets executed.
Anyhow, if you want to learn what's on the chip check out this section.
Programming the D-Wave chip is nothing but initialising the spin states of the qubits, it has nothing to do with classical transistor logic. And no, it is not alien technology, although apparently pretty foreign to you.
Yet, you write like somebody who already made up his mind, and I doubt you're willing to learn anything that contradicts your preconceived notion.
Well then they duped them more than once and also duped Phys. Rev. A and B. Phys. Rev. Lett.
http://www.dwavesys.com/en/publications.html
Sorry pal, but you may as well subscribe to creationism. They have an adiabatic chip, the only open question is how good it is.
I've been following the Rossi story as well and agree that it is a fraud, but the comparisons to the ecat are only superficial.
The contraptions Rossi builds are cheap and look like a plumber put them together. On the other hand D-Wave has chips samples on display that are produced by a special purpose foundry that can produce Niobium SC circuitry. That took some serious investments.
Rossi supposedly sold his house to finance his venture, D-Wave is backed by the likes of Steve Jurvetson and Jeff Bezos. Rossi has phantom customers who don't talk to the public, D-Wave has Google, NASA and Lockheed Martin.
Frankly, after they started publishing in Nature it's ludicrous to hang on to the idea that this is just an elaborate fraud. Then again conspiracy theories are a dim dozen on the Internet, and you are free to believe whatever you want.
Sorry to blow your bubble, but you can actually look at the chip that's inside D-Wave's boxes. It doesn't have transistors, it has Josephson junctions. Tell me how you can get any calculations out of those unless you do physical annealing with them.
Yes, the quantum bitcoin miner thought occurred to me too :-)
There are good reasons people were suspicious of D-Wave, the way they first made a splash and overpromised delivery pushed all the right buttons.
But to hang on to this stance after the amount of scrutiny that the D-Wave machine received is about as rational a climate change denial.
It's one thing to argue that they have not proven a quantum speed-up, but they clearly build an quantum annealing device that you can use to perform calculations.
I was referring to the paper by Mathias Troyer et. al. that is yet to be published not the effort that Alex Selby writes about (thanks for the link).
Will have to read the latter in more detail to get a good grasp of how much effort is required to beat the benchmark with Alex's approach. Two caveats: On first glance I am not sure if he has the same training data (he mentions he communicated with Cathy McGeoch) - if he does it'll be interesting to see how stable his generic approach is when the problem domain is slightly altered.
Do you know if he plans to publish any of this?
The problem is mostly on the engineering side, it'll be hard to build up industrial capabilities of the scope needed for a useful universal gate based QC and conceal if from the rest of the world. It's not like coming up with a code cracking algorithm, the latter could be kept classified quite easily.
No, "not in stealth mode" means they are now publishing like clockwork.
Enlightment is just one scholar.google.com search away.
That'll be a lot of effort for a fraud, especially since you then have to fake all the qubit specific data that goes into the publications. And the chip samples they have on display very much looks like Josephson junctions circuits and nothing like regular chips. (And the integration density they have for this process could not at all deliver reasonable classical performance).
Given the recent NSA revelations I think that our security in mostly illusional anyway, but if it makes you sleep better, the D-Wave machine cannot implement Shor's algorithm.
Ahem ... the central claim that their chip is a true quantum chip leveraging qubit entanglement has been demonstrated i.e. this paper.
Matthias Troyer, one of the co-authors, expressed in an email to me, that he was surprised to see this evidence, but that the chip seems indeed to perform some sort of quantum annealing.
The thing is operating close to absolute zero, and indeed the increase in noise with temperature severely degrades the machine's performance, but this is exactly as you'd expect, so I fail to see your point.
The secrecy pretty much went away when they came out stealth mode.
Now they published several papers in Nature and are quite open.
When I visited them I was surprised that there were no restrictions on taking pictures and nothing was off limits.
I recently visited D-Wave looked at their chips and deep freeze containment. Shot a snapshot of Geordie Rose standing in one of the open boxes.
You may think they are misguided, but their tech is for real. Even Scott Aaronson doesn't deny that.
There is no classical computer hidden inside, but there is still reasonable doubt as to exactly how quantum the device is, and if it will ever deliver clear cut quantum speed-up.
No, the D-Wave machine can only solve the Ising equation. Universal adiabatic quantum computers have been shown to be able to emulate gate model quantum algorithm, but for the more restricted current D-Wave architecture a mapping is (probably) not possible. Nevertheless the class of problems they can solve is still pretty large, and is applicable to useful optimization use cases and learning algorithms.
The benchmark did indeed not demonstrate a quantum speed-up, but it in fairness to D-Wave this was a test designed based on the customers requirements i.e. for them acing this benchmark was good enough to justify investing in this technology.
My understanding is that the algorithm that was comparatively fast on a classical computer was hand optimized by a graduate student, it was not a generic annealing algorithm solver.
But the paper on this effort of 'beating' D-Wave on a classical machine is yet to be published, so this is all from blog hearsay.
You are a bit behind the times. This was true as long as D-Wave was in stealth mode.
At this point they are quite open and have published several papers in Nature.
... I got the impression that he is not overtly concerned about this ongoing controversy, although he did mention he prepared another paper to demonstrate entanglement on the chip.
But his focus is clearly on tackling hard tasks with immediate applicability (for instance in deep learning).
... seems to be rather unimpressed with the idea
If it was only shoddy experiments that'll be one thing. But some theoretical physics has gotten so used to divorcing itself from experimental verification it's turning into fairytale science.
Maybe that'll throw a monkeywrench into the viral notion that we are living in a simulated universe.
Not holding my breath though.
Reputable journals are only marginally better. Just witnessed a back and forth where some research was attacked by a prominent scientist. The assumptions the latter made weren't quite on target, so the attacked researchers submitted a paper pointing this out. This passed anonymous peer review but then the paper solicited the opinion of this star scientists. He dismisses the paper with the most bizarre arguments that give the impression that he didn't even read it. Then the prestigious journal turns around and endorses this position and rejects the paper.
Absolute astounding and sobering.
Disclaimer: I am not a party or author in this but covered the controversy on my blog, and hence have been shown the rejection email.
Flat earthers, creationists and climate change deniers are immensely annoying but this is still a terrible decision. Trolls will never go away, but they are only kept in check by challenging them.
On the other hand science can benefit from intelligent and and well informed commentary and like any other human activity benefits from critical, constructive media coverage and public debate.
In the final analysis science is too important to just leave it to the scientists.
Second that.