Google Announced their D-Wave 2X Quantum Computer Works (2015)
popularmechanics.comAs far as I'm concerned, it doesn't work until Scott Aaronson says it works.
EDIT: My comment was kind of snarky and curt without much info; I need to be better about that. I did read Scott's post when it was published, and the consensus I got from it is that D-Wave's device achieves a constant-factor speedup, not a quantum one. I just think, with that being generally accepted by all parties now, it's pretty disingenuous to keep calling it a "quantum computer", when it's really a (faster) classical computer that uses quantum mechanical effects. I mean, the Intel chip in my laptop also uses QM effects because the feature size is so small, but nobody calls it a quantum computer. Maybe that's the media's fault though. Are D-Wave/Google themselves still saying "quantum computer"?
SECOND EDIT: Rereading Scott's post more carefully, it seems like Google and D-Wave are now calling it a "quantum annealing device" and are more forthcoming about the lack of quantum speedup. So unless they're talking out of both sides of their mouth and still saying "quantum computer" to the popular press to build hype, I guess everyone is a reasonable person after all and it's the media's fault as usual.
Last time I understood they got a classical speedup. And yeah if Scott says it ain't working, it ain't working, agreed there.
I am wondering if they demonstrated there is quantum entangling taking place between qubits and have they observed an asymptotic speedup?
Wonder if there is some hesitation to just say, yeah this doesn't work. We spent all this time and money on it, but that's ok. We learned what doesn't work that is still valuable.
I studied quantum computing a while back (maybe 7-8 years ago). There was a feeling it was going to be the next big thing. Grover's search and Shor's factoring was promising. But I think we ended up mostly with pictures of cats in the cloud and Javascript on the server ;-) which is ok too, I guess.
This article and the discussed work was put out last December (see http://arxiv.org/abs/1512.02206). Scott Aaronson weighed in around that time; see http://www.scottaaronson.com/blog/?p=2555.
The blog entry includes a reference to the HN discussion in the updates toward the end.
The previous HN discussion clarifies that google reports a constant (although large) increase in speed for a specific application.
I think Aaronson's Q&A w/ MIT News is the most accessible explanation of why this isn't an "it works!" moment. It's a complex, nuanced argument so it's easier to just read than try to summarize here [1]
I think basically if you grasp why Selby's algorithm is able to beat D-Wave, you'll understand why D-Wave isn't a useful quantum computer, whatever terminology you want to use to describe it.
N.B. Even Aaronson admits D-Wave may have useful technology[2], it's just that the present "dirty qbit" approach doesn't give you a machine useful for anything more than basic science research, which is not at all how D-Wave have hyped it.
[1] http://news.mit.edu/2015/3q-scott-aaronson-google-quantum-co...
[2] http://www.scottaaronson.com/blog/?p=2555#comment-967324
> We note that there exist heuristic classical algorithms that can solve most instances of Chimera structured problems in a timescale comparable to the D-Wave 2X. However, we believe that such solvers will become ineffective for the next generation of annealers currently being designed.
From the abstract.
I was wondering if these exist. My last counter of D-Wave is they could be using a good algorithm with specialized ASIC's or regular HPC hardware in a cluster. As in, they're certainly accelerating things but making up the quantum part. That possibility has to be eliminated.
That there's classical algorithms that might get a similar speedup, esp if on custom hardware, validates my prior concern.
I was about to post the very same quotation.
Its not naively clear where the line is drawn between "speedup that could be obtained through clever heuristic/nondeterministic/whatever classical algorithms" and "speedup that is necessarily the domain of quantum computation." I suspect the question is a deep (open?) one in computational complexity theory.
google shouldn't call anything wave :)
Just for the anonymous record... I came up with the design for Google Wave, it was a protocol I was working on. I won't name names but I met someone whose name rhymes with Cint Verf at a charity dinner and in a stupid attempt to impress him hoping he'd want to do business described every aspect of the exact product they came out with over a series of emails 8 months before Google published Wave.
Just to include me in the development somehow would have meant so much to me, instead I feel stabbed in the back by someone who was a hero to me. It's my fault because I foolishly told him and I didn't publish my code first but that's still how I feel.
What?
The name has nothing to do with Google. D-Wave Systems is the name of the company that developed and manufactured the computer.
I liked Google Wave... at least better than Google+
Same here!
RIP Google Wave
It lives on in Apache!
If by "lives on" you mean exists in a repo sitting on a hard disk somewhere, then yes, it lives on.
I haven't seen anyone do anything with it or for it since Apache got hold of it. Nothing against Apache, here, just that I think the market demand for this software has fallen off a cliff.
I thought it wasn't considered a quantum speedup if you're talking about a constant factor. Otherwise, how are you going to jump across complexity classes?
talk about changing goal posts and pivoting, anything to save face about rushing success.
quantum everything is obviously the future of technology, but it is a long long path.
i get more excited about fundmanetal quantum research and research on new methods of testing quantum states.
people claiming to have full working computers of any value, if they even do work-------------are just blowing smoke and mirrors like elizabeth holmes from theranos. nearly criminal behavior , alleging things are simply what they aren't in order to get money.
there is an epidemic of hack science and engineering occuring to secure and suck in scarce funds. and where-ever there is military related big money the likelihood of fraud is even bigger.
read the book about the hafnium bomb and you will understand why the military investors are frequently so easy to scam---a two pronged version of greed and desire to have the unstoppable weapons.
paradoxically enough , if more pre-investigation was done, they'd have more money to spend on real R&D. i guess economists would just call this a massive case of malinvestment.
I got about halfway through the article wondering why the NSA wanted to talk about their quantum computer so much, then realized it was NASA.
We changed the url from http://trendintech.com/2016/05/07/google-announced-their-d-w... to one of the (more) original articles about this, that didn't specifially get discussed last year.
There were a few discussions at the time (e.g. https://news.ycombinator.com/item?id=10698317) but perhaps the community is interested in more.