I went to the D-Wave demo of the "first commercially viable quantum computer." That's a heavily qualified statement, but it basically means that they were able to take a 16-qubit quantum computer and hook it up to a molecular structure search, a seating chart constraint optimizer, and a sudoku puzzle solver. The cool aspect of what they done is that their software is designed to optimize SQL, so you don't really need to know anything about the adiabatic quantum computation. Their software will take the SQL program, convert it to a graph, and load it into their quantum computer. The best solution is lowest energy state of the quantum computer. Due to the physical nature of the computation, the second-best solution is the second-lowest energy state, etc... Their goal with this sort of model is to either:
- Get a more accurate solution in the same amount of time as a digital computer
- Get a solution of the same accuracy but faster than a digital computer
Of course, their current prototype is about 100x slower than a current digital computer, so it will take awhile for them to transition from proof of concept into realizing those goals.
D-Wave's near-term vision for their technology seems to be an Internet service by which data can be sent to a quantum computer and a solution returned. This would in effect be a "quantum computer co-processor" for digital computers. They were very clear that quantum computing is not a replacement for digital computing; rather, it enables new efficiencies with respect to certain types of computation in finance, biology, chemistry, security, etc...
I went to the presentation to see the demo, but I was a bit let down by it. In retrospect, I should have expected to be: what they currently have is un-demoable. Their prototype is 100x slower than a digital computer, so it can't perform any type of new computation that would really knock our socks off (e.g. cracking a 1024-bit RSA key). Regardless, it is very difficult to demo computation. The only thing left to show would be the actual quantum computer, which in publicity photos looks absolutely beautiful. But there was no way they were going to transport a massive, liquid-helium-cooled quantum computer to the Computer History Museum. Given all this -- the no-show quantum computer, no demonstrable new application -- it pretty much leaves the rest of the demo to a question of belief. It would have been nice if they could have brought parts from previous prototypes, a mis-fabbed qubit, anything that could have taken their Internet-based demonstration and made it more physically grounded, but I complain because I like toys.
I've posted some videos in the extended of the molecular search demo. As I've already said, there's not much to see, just much to believe.
Molecular structure search part 1:
Molecular structure search part 2: