D-Wave’s Quantum Computers are the Real Deal
November 14th, 2007 by MarkI’ve been interested in the Vancouver-based startup D-Wave for over a year now. They make quantum computers. And, unlike any other company in the world, they’ve actually made working ones. This week they’re demoing a pattern recognition test that’s very difficult for traditional computers– scanning pictures and quickly identifying which ones are of faces.
The D-Wave Exhibit at the SC07 Supercomputing Show
“Our image-matching demonstration, the core of which is too difficult for traditional computers, can automatically extract information from photos–recognizing whether photos contain people, places or things, and then categorize them by visual similarity” said Geordie Rose, D-Wave founder and CEO.
Last year, Google acquired Neven Vision (an image-recognition company) for its expertise in recognizing similarities among photos–the simplest task being detecting whether or not a photo contains a person, and the most complex of which would enable accurate classification of images by person, place, and thing. Unfortunately, even after crafting the algorithms so that they sidestep the most difficult image-recognition problems, they were still too slow to be deployed by Google.
EE Times: D-Wave taps Google image search for quantum computer
The founder’s blog has quite a bit more information, as well as a picture of this chip:
Geordie Rose: A picture of the 28-qubit superconducting demo chip
I would say the company is making great progress. They’ve managed to scale up from 2-qbits last year to 16-qbits this spring to 28-qbits as of this week. According to Rose, they are on track to have a 500-qbit computer by next year. If they can do that, their quantum computers will be more powerful than any supercomputer on the planet for certain problem spaces, including nano-chemical engineering models, pharmaceutical research and cryptography. That last one worries me a bit. I’m invested in D-Wave (via Harris & Harris), and if governments realize how quantum processors scale up, I can’t imagine them not getting involved in some way. This technology will change the world.
Related articles
Business Wire: World’s First 28 qubit Quantum Computer Demonstrated
The Register: D-Wave qubits in the era of Quantum Computing
Engadget: NASA backs quantum computing demonstration
November 14th, 2007 at 5:21 am
Well since you put my name in the ‘Tags’, I’ve got to at least say something! So here we go…
Last I had heard, it was unclear whether quantum computers can easily scale. There’s a problem with keeping the qubits stable for the duration of the calculation. As you add more qubits, the problem gets worse.
The other thing I heard was that you need A LOT of qubits to solve meaningful cryptographic problems. I believe that even with 500 qubits, a cluster of computers (for a fraction of the cost) can still out-perform the quantum computer for its proficient calculations.
As an investment option, I would be very hesitant to go with D-Wave, but if they are able to scale the qubits gracefully, the potential is very large. Maybe a 90% chance they go bust, but a 10% chance they give a decent return. In any case, it’s cool technology to follow, and I’m happy to see it get funding.
November 14th, 2007 at 8:16 am
Actually, D-wave’s Geordie Rose has said repeatedly that the type of Quantum computer that they’re currently building is not suitable to break encryption. Although their adiabatic approach can be extended to be computationally equivalent to the gate-model approach, this seems to be not planned at the moment.
November 14th, 2007 at 1:57 pm
512-qbits can do quite a bit in terms of encryption breaking and 1024-qbits would far, far outstrip any other computer on the planet.
Dr. Rose said on his blog that they should be able to factor RSA-704 once they hit 512-qbits. Read the 12th and 13th comments to the post on his blog that I cited for the picture of the chip.
November 14th, 2007 at 8:53 pm
I was under the impression (from reading some of Schneier’s stuff) that quantum computing made it expontentially easier to solve certain types of factoring problems (ie. breaking RSA), but would not instantly obliviate other sorts of encryption algorithms.
The potential for these devices for pattern recognition and neural networks is very cool.
December 28th, 2007 at 4:15 am
I have a bunch of data and photos that I encrypted with
DH/DSS 1024/2048 and I forgot my passphrase. So I’m
rooting for DWave so I can get my stuff back!
I bet they’ll have a 1024 qbit prototype in a less than
3 years!
May 12th, 2008 at 3:58 am
I have become somewhat concerned by the D-Wave blog’s frequent suggestions that a quantum computer would revolutionize the calculation of NP-complete problems. From what I understand, q-bits allow factorization (e.g. RSA breaking) in polynomial time, but NP-complete problems are a completely different beast. Most likely a quantum computer would only reduce the time to square root of that required by a conventional one. For a 1000 point traveling salesman problem, the a quantum computer would require 2500 steps, rather than 21000. In other words, NP-complete problems would still be essentially unsolvable.
http://www.scottaaronson.com/writings/limitsqc-draft.pdf
I’m still convinced that D-Wave has a true break-through, but these kinds of unsubstantiated claims are a pretty big red flag.