World's Largest Quantum Computation Uses 84 Qubits | Hacker News Reader

World's Largest Quantum Computation Uses 84 Qubits

technologyreview.com

34 pointsllambda12 comments

Sure wish I could still flag submissions.

As maggit points out, (exactly as I did 228 days ago[1]) D-Wave has a long, long history of writing press releases that sound very interesting; then completely failing to follow them up.

An Arxiv preprint is a step up from a press release, but it's still not a peer reviewed paper. I haven't read it, nor do I know enough number theory to comment intelligently on its contents, but judging from D-Wave's past history of behavior, this paper is almost certainly fraudulent.

Extraordinary claims require extraordinary evidence. This is not extraordinary evidence.

1: http://news.ycombinator.com/item?id=2593323

You can purchase a working D-Wave machine today. I'm not sure how that qualifies as "failing to follow them up".

Easy, I'll quote the end of the article.

The company's approach to computing, known as adiabatic quantum computing, has been heavily criticised. Various physicists have said that the theory behind the machine is flawed and that the company has shown little evidence of the kind of improvement expected over classical computers.

Scott Aaronson has blogged about the misleading claims of D-wave in the past. Here is his take on the Nature article they published back in May, which he found to be their least bad effort to date, i.e. the only one that he wasn't able to immediately discredit as a computer scientist.

http://www.scottaaronson.com/blog/?p=639

There seems to be some controversy as to whether or not the D-Wave machines actually are quantum computers. I don't understand this fully myself (so please elaborate if you can!), but it seems incredible that this 84 qubit quantum computer would just pop out of nowhere when the state of the art seems to be just a few qubits.

http://en.wikipedia.org/wiki/D-Wave_Systems#Criticism

I'm also not sure whether it can be called a "classical" quantum computer.

On the other hand, the large corporate and government interest probably means that the devices are at least useful (or promising) for some kinds of computations.

Either that or it's a very, very shrewd con.

> just pop out of nowhere

That's how quantum computing works.

My very limited understanding is that they have built a kind of analog computer that solves minimization problems. They claim it's an "adiabatic quantum computer" (a relatively-recently theorized implementation of quantum computing) but the skeptics think that it works by some classical method such as simulated annealing and therefore will never outperform classical computers.

Can somebody elaborate? I am waiting for the day I get to wet myself with glee and shamelessly create arbitrarily complex Bayesian statistical models and brute force the parameter estimates with MCMC on a quantum PC... that will change things.

A good friend of mine with a PhD in this stuff points out that to do any useful work (ie breaking 3DES) you need a qubit for every bit in the factor(key), and all must be super-positioned at the same time - and that getting each extra bit superpositioned seems to be getting exponentially harder - in short quantum computing is looking to be technically much much harder than hoped. Kind of next century harder.

Yes, but I'd bet that lots of real world cryptosystems have bugs in them that reduce their effective entropy to 84 bits or less.

I would like it to be true, but apart from computing known values of Ramsey numbers, they should have computed unknown values of Ramsey numbers, to make their research more valuable.

As the bounds of small Ramsey numbers are known, it would provide more evidence of correctness of their computation.

Reply on news.ycombinator.com