February 23, 2006

Counterfactual computation

Posted by Arcane Gazebo at February 23, 2006 5:18 PM

I've been neglecting the blog the last few days, in favor of things like data analysis. Although I might have preferred to be doing things like Half-Life 2 instead, the data came out very well, and you will certainly see it if you attend my March Meeting talk.

In other quantum computing news, a group at UIUC has performed a very interesting experiment in which they combined quantum computing and quantum interrogation to get the result of a quantum algorithm without actually running it. (Via all over the place.) So at least one person will have a March Meeting talk that's much cooler than mine—for us "counterfactual computation" is when our qubits don't work—but in the spirit of quantum oneupsmanship I will note that my qubits are (allegedly) scalable.

UPDATE: John Holbo speculates about technological advances that may follow from this.

Tags: Physics, Quantum Information, Science

Wow, that's like watching the home video for a movie before the movie's even over.

I'll be giving a fun talk at the March Meeting. It's about Congress... The "Zell Miller plot" will be shown. :) While I don't think I have much of a chance of one-upping the UIUC guys in terms of the importance of the work on which I'm speaking, I bet my talk will be more fun.

Posted by: Mason | February 23, 2006 8:20 PM

FWIW, saying you've got a scalable qubit is about the only thing that'll get my rocks off--that's the only way to cross the threshold to something genuinely new.

Posted by: Lemming | February 24, 2006 1:37 AM

Oh, and I suspect my understanding of what's going on is way short to say this with any certainty, but I strongly feel that the real news here isn't that the program worked without being ran, but our notion of what it means for the program to "run" is inappropriate in this context.

Sure, the whole thing is cool (though the "cannot be scaled up" gives me ... dysfunction faster than... I'll stop there) it just kinda frustrates me. It's just that the productions that result in the correct information at the end don't really correlate to the traditional channels we think of the computation happening on.

Am I just nuts? Stupid? Sexy?

The program ran, it's just that the computer isn't what you think it is, and even the forward direction of propogation of information isn't necessarily what you think it is (well, I'm not sure wrt this problem in particular)

Posted by: Lemming | February 24, 2006 1:47 AM

Lemming: Your complaints seem very reasonable to me. Certainly, implementing something that can help reduce errors would need to happen in something scalable to have a practical impact given that these things will have to be scaled to actually do computations with these things.

To me, the excitement comes from the fundamental physics level rather than potential practical impact. One key will be to try to implement something like this in the potentially scalable situations.

Perhaps your different reaction speaks a bit of the CS perspective on this business versus the physics perspective?

Posted by: Mason | February 24, 2006 10:13 AM

I think Lemming is also getting at a point about interpretation of quantum mechanics. It's fun to talk about "counterfactual computing" but this is sort of a Copenhagen way of thinking about things. In practice I tend to think in terms of wavefunctions-for-everything, where (I think) you would find that there's a part of the wavefunction where the computation does run, it just doesn't interact with the part that we live in.

At some point I realized that the way I think about QM is basically just the many-worlds interpretation with the metaphysical language toned down. I'm reluctant to talk about multiple universes when they seem to me just to be other parts of the same world. I'm sure I'll change my tune when I open a portal to another world down in our qubit test chamber, and headcrabs start pouring out.

Hmm, maybe I should do a post on QM interpretations. I'd have to do some reading in order not to sound like an idiot, though.

Posted by: Arcane Gazebo | February 24, 2006 1:19 PM

I think your right. Really I'm only whining about semantics, not the awesomeness of the work (aside from my permagripe about scalability).

What would really blow my mind would be if it was possible, given the rules of our universe, to really build a nondeterministic computer. Until the scalability issue is fixed, which isn't just an engineering question IMHO, any implementation of a quantum computer is effecitvely limited to a fixed ceiling on its number of states, which in turn is not different in it's capabilities (time-wise) from a deterministic computer with a larger "state". Sure, it might be a constant factor faster, but if the number of states are bounded, it's not really a nondeterministic computer.

So, do the rules of our universe actually allow this sort of computation to happen? Is this a theoretic or practical question? I'd actually say theory, but it's a bit fuzzy (SCSI!), no?

Posted by: Lemming | February 24, 2006 1:20 PM

Async posting goodness! My reply was to Mason's post, just to be clear. And thank you AG, I think you expressed it better than I could.

Also--I suspect now I know what keeps you motivated to stay in shape with all this silly running. You expect to be in Gordon Freeman's shoes some day, and it's probably easier to just be well-prepared to GTFO than pick up a crowbar and start swinging.

Gordon Freeman didn't have New Balance.

Posted by: Lemming | February 24, 2006 1:25 PM

Not completely sure what y'all are talking about but Googling "Gordon Freeman" and "headcrabs" proved to be somewhat helpful to following the discussion.

Posted by: JSpur | February 24, 2006 1:30 PM

Yes, there are indeed fundamental physics issues with scalability even before the engineering. Excitement for new results notwithstanding, I think this is appreciated in the quantum computation field nowadays. (Though I did once run into an undergrad [just graduated, actually] at a conference who was attempting the semantic 'quantum teleporation has been done bit' and wasn't interested in issues such as number of qubits involved. He didn't seem to get it, so I pretty quickly dismissed him as somebody to whom I should pay attention. Then he spent the rest of the conference hitting on some (married) postdocs I know.

Gazebo: I was eating lunch with some the theoretical cond-mat crowd today and your advisor's name came up. I sadly didn't catch the details (and the conversation moved on before I could find out more), but I got the distinct impression they are not so impressed with his lab's work in general. Anyway, you probably know much better how some of the theoretical crowd is reacting to his work. (I can name names privately if you want.) Not that it even matters what a tiny subset of the theoretical crowd thinks. I'm just dutifully carrying out my spying duties. (First theoretical condensed matter physics, then the world!)

They also told a story about the T's effect on lab measurements in Boston. (How come we can only get good measurements between 2am and 6am...?) It reminded me of your comments about the noise you get in your building.

Posted by: Mason | February 24, 2006 2:11 PM

Historically some experiments in this building have been affected by the BART, but ours seems to be immune to that at least.

Posted by: Arcane Gazebo | February 24, 2006 4:11 PM

Testing my ability to comment wirelessly. Gazebo feel free to take this comment down if it actually goes through.

Posted by: jspur | February 25, 2006 4:56 PM

I'll leave it up as a reminder that I need to hurry up and post something for people to comment on.

Posted by: Arcane Gazebo | February 25, 2006 4:59 PM
Post a comment