Comments on: Thanksgiving Special: D-Wave at MIT

By: Jonathan Vos Post

Jonathan Vos Post — Wed, 28 Nov 2007 16:06:30 +0000

Analogies are very important in Science and Mathematics, just as a paper without a “narrative” is harder to get emotionally involved with.

In the very old days, I think that I could have convinced Euclid (if I spoke Greek) and much later, Newton, that in Euclidean space:

“An analogy is a parallelogram in semantic space. To say A is to B as C is to D, where A, B, C, D are statements about the physical universe, social universe, or an abstraction, can be translated.

The line segment jointing A to B is parallel to and the same length as the line segment joining C to D.”

This generalizes to vector spaces, or bimodules for noncommutative cases, and further to metric spaces. One actually sees this done, sort of, in Web 2.0 pages that analyze web pages and make 2-D or 3-D maps of the keywords or sub-domains based on a metric of distance.

In that Euclidean sense of analogies, what is an analogy between analogies? Is it a 3-D parallelopiped? How does that generalize?

By: Michael J. Biercuk

Michael J. Biercuk — Wed, 28 Nov 2007 14:53:27 +0000

I disagree. While a detailed technical discussion among an audience entirely composed of experts should not use analogies, the breadth of topics in question suggests that analogies may have a useful role in highlighting how the pieces fit together in this discussion. Analogies serve to assist in understanding general concepts, and their use should be limited to such purposes, as is done here.

By: Herb Ivorous

Herb Ivorous — Wed, 28 Nov 2007 13:24:45 +0000

Analogies and metaphors can add to confusion. For example, physicists’ analogy of a rubber sheet for spacetime is misleading. They show a ball depressing a rubber sheet and say that gravitation is the result of the sheet’s shape. This is not conducive to understanding. The omelette analogy is also a mistake.

By: Michael J. Biercuk

Michael J. Biercuk — Wed, 28 Nov 2007 13:17:46 +0000

Oh Tyler…

Greg - sorry for not being more artful. I don’t really have a significant ability in that area. Note that I didn’t say I was refining your analogy - just revising it. But also note that the nanny in my version isn’t just supporting classical hardware/software. It’s a completely parallel classical computational path using the superconducting hardware - thermal annealing. And it may have actually done what little work was attributed to the “quantum computer.”

By: Tyler DiPietro

Tyler DiPietro — Wed, 28 Nov 2007 06:10:23 +0000

I’m surprised that with the omelette analogy being fully solidified in this thread, no one has made any jokes about how D-Wave might be thinking that you can’t make an omelette without breaking a few eggs. Fill in the details.

By: Greg Kuperberg

Greg Kuperberg — Wed, 28 Nov 2007 05:52:20 +0000

Michael: I get what you are saying now. Maybe you could more artful with your analogy, especially since it came from my analogy.

Let us say, more simply, that D-Wave boasted that they have a 2-year-old child prodigy who can make an omelette. (That is, solve a Sudoku.) But the omelette was actually made by a house cook who only asked the “prodigy” to fetch the eggs and onions. They insist that the toddler has special mental powers even he needs a lot of help in the kitchen, but actually he seems entirely ordinary. (That is, non-quantum.)

Now another question is whether there actually was a nanny (meaning other supporting hardware) who stepped in for the toddler to fetch the eggs and onions. Or if on the day of the demo, the house cook didn’t bother talking to the toddler at all because it was too much trouble. Varun Jain did what he could to make the story sound more honest, but his side of it only goes so far.

By: Michael J. Biercuk

Michael J. Biercuk — Tue, 27 Nov 2007 21:33:33 +0000

I think the design rules would indeed be useful for this simple algorithmic implementation. However, on the hardware side, we know roughly how they make their devices based on published papers in PRL, and the JPL fab process. However, we do not have a clear understanding of the “quantumness” of operations on these devices. DWave has produced an interesting, but thus-far unconvincing theory to suggest why operations on their system are quantum. However, a demonstration of algorithmic performance plus a theoretical explanation of why the behavior could be quantum is startlingly insufficient. Instead, definitive experimental evidence that the devices are performing quantum rather than classical analog operations is required (e.g. Chuang or Bacon’s ideas on varying temperature). Otherwise, aside from being disingenuous at best, it is impossible to say if such devices have any chance of working in more complex forms.

By: John Sidles

John Sidles — Tue, 27 Nov 2007 18:15:53 +0000

Michael, I think all the posters on this thread would agree that even a sketchy description of D-Wave’s design rules (both hardware and algorithm-related) would be very welcome.

Here “sketchy” means, “the minimum sufficient to suggest concrete avenues for research.” Most technology companies find ways to do this … mightn’t D-Wave manage it too?

By: Michael J. Biercuk

Michael J. Biercuk — Tue, 27 Nov 2007 17:13:11 +0000

Hi Greg,
I’m saying two things - the classical controller did the majority of the work, and what was left for the “quantum” part may very well have been achieved in a purely classical way. The nanny is the classical process which may masquerade in place of the quantum part. “You”, as the true omelette chef are the classical controller performing the bulk of the work, while the true contribution from the four-year-old is very difficult to discern.

By: Greg Kuperberg

Greg Kuperberg — Tue, 27 Nov 2007 16:15:40 +0000

So, Michael, what are you saying about the 16 qubits? That they didn’t show much quantum behavior? Or are you saying that their classical controller — presumably this “nanny” that you describe — did MIS for 6-vertex graphs instead of having the 16 qubits do it?