Quantized Thoughts

"A TP from Oxford who gets riled by the Steorn debacle. Some other topics too, but that's what stands out for me." - TCD Mathsoc Wiki

Wednesday, December 19, 2007

QEC highlights so far

I've been sitting here for the last three days watching talk after talk on quantum error correction and fault-tolerance. I don't think I've ever been to a conference before where the material was so consistently interesting (to me at least).

I gave my talk yesterday, but laptop problems the night before meant that I missed four of the morning talks while trying to reconstruct my slides.

There have been a couple of talks which really caught my attention over the last few days, and I thought I'd mention them here.

On Monday there was a set of four, hour long, tutorials, given by Daniel Gottesman, Lorenza Viola, Dave Bacon and Ray Laflamme. I really enjoyed each of these, and I think they worked well to put everybody on the same page before the invited talks and contributed talks.

Debbie Leung gave a talk on measurement errors in the cluster state model, and showed that they are trivially equivalent to Pauli errors in the circuit model if we choose a specific interpretation of what happens in MBQC. While the trick is very simple, I really like it, and enjoyed the talk.

The other talk which caught my attention that day was given by Mohammad Amin from DWave Systems. In the talk he argued that their adiabatic quantum computer was robust against multiqubit dephasing (assuming this is in the energy basis, I have no problem with this) but also that it is not limited by single qubit decoherence times. This strikes me as highly unlikely. Actually, I think it must be impossible. He argued that we only need to consider the lower lying energy levels when talking about decoherence in AQC, and proposed a two state model. My problem with this is that because of the non-local nature of the energy eigenstates near the anti-crossing, local errors should affect all energy levels. The slides from his talk are available online here. I haven't seen him around at coffee, but I'll try to ask him why he doesn't consider this to be a problem if I see him.

The talk the next day which has stirred up the most controversy was given by Robert Alicki arguing the impossibility of fault-tolerant quantum computing. There was quite a lot in the talk that I found strange, and actually thought a lot of the arguements were orthogonal to the way we actually go about quantum error correction. That said, there were a few parts where I didn't follow his argument, so I'm probably not in such a good position to comment.

Dave Bacon is also blogging about the conference and has mentioned the latter two talks here and here. The second link is an open thread on the conference.

Also, yesterday I gave my talk on globally controlled fault-tolerant quantum computation. It seemed to go down pretty well, but I got a little rushed at the end when I saw the session chair holding up a 1 minute sign. The poster session contained some very interesting results too, and maybe I'll get around to talking about those in another post. Michael Bremner's poster on noise in globally controlled schemes certainly caught my attention, since it relates so closely to my work. Actually I think we've seen the effects of it in some recent NMR experiments. I'll write something about that later too if I get the chance.

Todays highlight, for me at least, was Panos Aliferis's talk on error correcting the IBM superconducting qubit. It seems that IBM are now below the threshold for fault-tolerance, which really is big news. It seems that they are just under the limit, so the overhead will likely be large, but even so, this really is a major achievement. The talk was very clear and easy to follow, but unfortunately, there were a lot of numbers for the error rates which I didn't manage to scribble down. Hopefully the slides will be available on line after the conference. Anyway, the over all message to take away from his talk is that asymmetry in noise can be a very good thing, and that we shouldn't obsess over CNOT fidelities if we have good controlled phase gates (even if we don't have good Hadamard gates).

Anyway, there is still half the conference left to go. I'm looking forward to hearing more results. There are a few talks I am particularly interested in, but more than that I'm looking forward to hearing results that I hadn't expected. It's always good to be pleasantly surprised by the state of the field.

UPDATE: It seems the numbers for the IBM qubit were estimates and do not relate to the actual experimental fidelities currently being achieved. This makes a lot more sense to me. Basically I didn't pay to much attention to the first few minutes of the talk (and hence missed the fact that they were estimates) and only really paid attention when the fault-tolerance discussion started. There have been a lot of very interesting talks, and so occasionally my attention drifts.


At 12/19/2007 10:27:00 p.m., Anonymous mick said...

Joe - you easily win the award for the best Simpsons analogy in a talk so far at this conference.

Thanks for the plug for my poster. I'm still not convinced that my noise model makes a difference to your scheme.

I liked Preskill's talk more than Panos's - both were awesome though!

At 12/19/2007 10:40:00 p.m., Blogger Joe said...

I don't really think it affects our scheme either, but its still interesting, since the underlying system is so similar.

At 12/20/2007 06:21:00 a.m., Anonymous Anonymous said...

Dwave don't want to buy patent of conventional computers design technology and so thus saying that they computer is quantum rather than classical.


Post a Comment

<< Home