A Strange Computer Promises Great Speed:D-Wave’s quantum computer

[ErikTheNorwegian]

A S

[Gall]

No imformation so far. How many qubits?

[FenderBender]

When will consumers see one of these?

[jpb]

My understanding is that the D-wave computer does not have qubits in the way that other (mainly theoretical) quantum computers have, instead it is more like an analogue computer in the sense that it is set up to solve a particular problem.

That is it doesn't use quantum gates so though the computer is described as 128 qubits they are not connected by quantum gates.

DWAVE is quite controversial :

http://spectrum.ieee.org/computing/hardware/loser-dwave-does-not-quantum-compute

but that article is about 3 years old and I think the blog author has become more positive about whether or not quantum effects are happening though perhaps not about some of DWaves more wild claims.

I'm interested in this as I help teach undergraduates quantum computation but this is of the quantum gate sort. Initially I was pleased that there appeared to be a real commercial quantum computer to talk about, but even if it does quantum compute it is completely different from the computation models that are the basis of the famous quantum algorithms that will allow rapid number factorization and code cracking.

[ejeffrey]

D-wave has gotten *much* better about their claims and the information they provide over the past few years.  They are no longer calling it a quantum computer (which in the scientific community has a specific meaning), but a quantum annealer, and they aren't making the same grandiose claims about what it does.  The easiest way to understand this is to read about simulated annealing: this is a classical computer algorithm to solve global optimization problems that works by simulating the classical behavior of annealing (where you cool something through the freezing point slowly so that no defects are frozen in).  The D-wave quantum annealer is a physical implementation of annealing, and it does so with a quantum phase transition rather than a classical phase transition.

Anyway, they have now showed some data showing that their annealer solves some problems faster than a single CPU core running commercial global optimization software.  The data they have shown are very limited, and of course they have a lot more classical computing power running the annealer than a single CPU.  Still, that is a big step and they are also now allowing people outside the company to test their hardware rather than suing people who doubt their claims.  Hopefully soon we will see some more data on the subject.

I am still reasonably skeptical about how valuable their device is going to be in a commercial sense, but I am really happy that they are now acting responsibly and trying to let data speak for itself.