...Independent quantum computing researchers said they are dubious of some of the claims made by D-Wave Systems Inc. because the privately held Canadian company has not yet submitted its findings for peer review, a standard step for gaining acceptance in scientific circles.
Many scientists believe that true quantum computing - which is based on the unusual properties of quantum physics - promises to solve certain factoring, simulation and other intensive problems faster than today's machines that rely on classical physics. Most say it's likely still years or decades away.
"Until we see more actual measurements, it's hard to know whether they succeeded or not," said Phil Kuekes, a computer architect in the Quantum Science Research Group at Hewlett-Packard Co.'s HP Labs.
Maybe they were premature. We'll see. ;)
I'm curious what you think about Scott Aaronson's thoughts on the subject.
ReplyDeleteDamnit, Carlos. I was supposed to be helping my wife with her biology last night - which I did anyways - and getting some sleep - which I didn't, in part because I sat there reading and chewing on Scott's commentary.
ReplyDeleteHe has some VERY (with blink tag) valid points about the Dwave system. The onus is on D Wave to prove they have a working system. That Geordie et al are being cagey is not a good sign.
Scott's theoretic underpinnings are to be given heavy weight, but, honestly, we need to be careful about theorizing about hardware that's not been built as yet. Dogmatic positions tend to be dangerous. There was a lot of dogmatism wrt to the vector vs MPP positions back in the day. While the MPP folk tended to be more right than wrong, there were a lot of dead bodies from both sides dogmatics in the end.
However, I will say that, rendering down the PR crap[1], this is largely what was expected amoung the HPC types first. Coprocessors, especially task specific ones, are easier to do than general processors.
I'm tempted to see if I could get the D Wave guys to come down for a visit and have them present. Dave Bailey, Bill Kramer, etc. et al would eat them alive if they're bullshitting.
Anyways, long story short, Scott's more on target than not. I'd be more cautious than he is though.
1. 'capability class supercomputer', frex. That has particular meaning to HPC folks, btw. Capability vs capacity is a HUGE war for us. Semantics and who gets tarred with them normally are in a particular field, alas. Rendering down from PR crap is something that you have to do for EVERY HPC press release. Alas, ours included. The longer term the player, the less the hype, but every single new kid on the block wants to claim (and inevitably does) that they are the best damned thing since Seymour. The overhyping is frakin annoying.
Least I could do. :-o
ReplyDelete