10

I've seen a lot of excitement in the popular press about the computers made by D-Wave Systems, but when I dig deep the only practical things that I can figure out that one can do with the computer are variations of:

  • Train one's staff in the use of quantum computers.
  • Write papers about what you can do with a quantum computer.
  • Explore quantum algorithms.

That is, these aren't even research machines---they are exploratory machines for what kind of research we would do if we had larger machines.

Am I mistaken? Is there any "quantum advantage" to be had from these machines---that is, a computation that can be done more cost effectively than with conventional computers?

vy32
  • 641
  • 3
  • 13

3 Answers3

3

There is a recent article that just came out couple of months ago related to Dwave's new chip. Even here, you can see that they mentioned that there is no "quantum advantage" over classical thus far on any of the problems they have considered.

I like the comment made by Malibu Stacey in the comment section... and I quote

It's really not. Is D-Wave's new hardware "faster" than using a "classical" solution like Fujitsu's Digital Annealer hardware?
Well you don't need to do multiple runs of the same problem on the Digital Annealer to have any confidence in your results like you do on D-Wave's hardware so no it's not any faster. Is it more cost effective? The Digital Annealer is just a piece of regular silicon (AKA an ASIC) which doesn't need to be be supercooled so it's far cheaper to run (it's built on a PCI-E card so you could plug it into any regular desktop assuming you could get Fujitsu to sell you one). Can it solve more complex problems? D-Wave's new hardware more than doubles the number of variables you can use to 5000 compared to their previous iteration but the first iteration of the Digital Annealer can handle a maximum of 8192 variables.

So it's a pretty simple answer. It's not faster, cheaper or "better" than the purely classical solution which has been publicly available for almost 2 years now.

KAJ226
  • 13,822
  • 2
  • 10
  • 30
1

It is true that quantum computing is still emerging technology, howvever, it seems that D-Wave will come to practical usage sooner than universal quantum computers such as IBM Quantum Experience. The reason is that D-Wave provides simpler single purpose computers focusing mainly on optimization.

See for example these links for near to practice application of D-Wave:

  • Lockheed Martin application of D-Wave for source code verification and mistakes identification
  • Volksvagen used D-Wave for traffic flow optimization in Beijing

Other applications seems more or less experimental, for example NASA partnership with D-Wave.

What is interesting is a usage of D-Wave by Los Alamos National Laboratory. It seems that army research is interested in the quantum annealers which can be significant for further development (mainly in terms of financing).

Martin Vesely
  • 13,891
  • 4
  • 28
  • 65
0

If there is any quantum advantage, this Lockheed Martin paper must show a clue.

In this paper, they utilized the D-Wave machine to sample the Boltzmann distribution and thus trained a deep neural network. According to their observation, their machine can sample far better than a classical computer at the same sampling times(I have not read their code for the classical method, so this statement might have some problem). But the author also committed that they have not provided any theoretical analysis. I myself is also trying to figure out that, or in a more rigorous way, if there are any quantum advantage.

Here is two of their plot.

enter image description here

Yitian Wang
  • 958
  • 4
  • 16