Total Pageviews

Wednesday, May 8, 2013

A Quantum Computer Aces Its Test

The long-sought quantum computer, a machine potentially far ahead of today’s best supercomputers, is almost as hard to define as it is to build. For at least a few particular uses, however, the unusual computer made by D-Wave Systems now seems to be very fast indeed.

Next week a professor at Amherst College will present her findings about the performance of the D-Wave machine, which its makers say makes use of such unusual properties of quantum physics as a particle’s ability to move in one direction and its opposite at the same time.

The professor, Catherine C. McGeoch, who is the Beitzel professor in technology and society at Amherst, gave the machine a so-called optimization problem and compared the results with those generated by  popular software from I.B.M. running on a high-performance machine.

The D-Wave machine, she said, was 3,600 times as fast as the  conventional system.

“There is no sense in which this is the definitive statement about quantum computing,” Ms. McGeoch said. “I’m more interested in how well it works, not whether or not it is quantum.”

That question matters a great deal to some others in the field. While quantum properties are among the most tested and proven domains of physics, the concepts behind them â€" for example, suggestions that we live in one of many universes, or that objects not in direct contact can affect each other â€" make such properties hard to accept.

Harnessing them for the sake of computation, suggested as a possibility more than two decades ago, has proved difficult.

The optimization problem is typically something like how a traveling salesman would plan a complicated trip most effectively. Ms. McGeoch tested three problems involving optimization. In two of them, the D-Wave computer was slightly faster. In the third, it was markedly faster.

D-Wave, which was the subject of an article in The New York Times in March, has been criticized for making claims about its quantum capabilities that cannot be supported.

Over time, however, D-Wave’s performance has improved, and the skeptics have toned down their criticism. Nonetheless, D-Wave is sensitive about the issue and, even after selling a working machine to Lockheed Martin, eager to rebut the criticism.

Ms. McGeoch, who has spent more than 25 years testing computer speeds, performed the experiments while on sabbatical and was retained by D-Wave to run the tests.

D-Wave solves optimization problems by setting them in the context of energy consumption: the lowest power needed to achieve a stated outcome, which it says is quickly achieved through a quantum process, is the answer. D-Wave thinks that many problems in computing might be restated as optimization problems and that its machine could be coupled with cloud computing systems for particularly hard problems.

Ms. McGeoch said D-Wave’s chips had performed well and might have better outcomes in the future, as its machines become more powerful, and more complex optimization problems are set.

“There could be a tipping point,” she said. “If the problems get big enough, conventional systems break down. In theory, you could solve a large number of optimization problems. People don’t know how to do that conventionally without losing a lot of efficiency.”