Evidence Emerges That Google’s Quantum Computer May Not Be Quantum After All
When Google paid a reported $15 million for a quantum computer last year, many physicists were skeptical.
Back in 2011, the aerospace giant Lockheed Martin paid a cool $10 million for the world’s first commercial quantum computer from a Canadian start up called D-Wave Systems. In May last year, Google and NASA followed suit, together buying a second generation device for about $15 million with Lockheed upgrading its own machine for a further $10 million.
These purchases marked the start of a new era for quantum computation. Theoretical physicists and computer scientists have been predicting for 30 years that quantum computers can dramatically outperform the conventional variety. And in May last year, these dreams at last appeared to be coming true when Cathy McGeoch at Amherst College in Massachusetts said she’d clocked the D-Wave device solving a certain class of problem some 3600 times faster than a conventional computer.
This finally backed up D-Wave’s long-pronounced but never confirmed, claims that their device was indeed faster than anything else at some tasks. Never had quantum computing’s stock ridden so high.
Since then, quantum computing—at least, D-Wave’s version of it— has undergone a dramatic change in fortune. And that culminates today with a report from a team of physicists from IBM’s T J Watson Research Laboratory in Yorktown Heights, NY, and the University of California Berkeley, who say that D-Wave’s machine may not be quantum at all. Indeed, its results could be just as easily explained if it was entirely classical.
First some background. Following McGeoch’s widely reported result last year, the first signs of trouble for D-Wave came from a group at the University of Southern California who said that the speed up she found was simply due to the the software she’d used for the comparison. Indeed, they rammed home the point by writing a piece of code for an ordinary laptop that matched the speed of D-Wave’s device.
Then came other evidence that D-Wave’s machine might not offer the computational benefits that quantum computers had always promised. The problem with conventional computers is that as problems increase in complexity, the time needed to solve them increases exponentially. That’s the basis on which many encryption schemes work—it’s not that they are impossible to crack, but that they are so complex that ordinary computers would take the lifetime of the universe to do it.
The big advantage of quantum computers is that they do not suffer from this problem of exponential time scales. That’s why governments and the military so fear their development—the first quantum computer capable of cracking these codes will make them obsolete overnight.
But last year, evidence began to emerge that the time D-Wave’s device needs to solve a problem scales in exactly the same way as an ordinary computer. The special quantum speed up appeared to be absent.
Despite these damaging results, D-Wave still had an important ace up its sleeve. One criticism aimed at its machine is that it is actually a classical computing device in disguise. In other words, it lacks the fundamental property that it hopes to exploit—it is not quantum in nature.
But last year, an international team of physicists showed that D-Wave’s machine was indeed a quantum machine, that it relied on quantum processes to achieve its results, whatever their speed.
This kind of evidence is actually harder to gather than it sounds, not least because any direct measurement of a quantum process automatically destroys it.
So physicists have to take an indirect approach. They assume the quantum computer is a black box in which they can input data and receive an output. Given this input and output, the question they studied was whether this computing behaviour could be best reproduced by a classical or a quantum algorithm.
Their conclusion was that the D-Wave machine was quantum in nature, that its results were closest to those from a process known as quantum annealing rather than the classical algorithms they also tested.
Quantum annealing is exactly the quantum process that D-Wave says it exploits so this result was a huge relief. The company’s team of scientists could breathe freely again.
That may change after a close study of the work from IBM and UC Berkeley. These guys, who include Umesh Vazirani, one of quantum computing’s early pioneers, point out that the black box experiment is highly sensitive to the choice of classical algorithms that the “quantumness” is compared against. Sure, the result shows that quantum annealing is a better explanation than the classical algorithms it was compared against. But perhaps there are other classical algorithms that do better.
Indeed there are. Vazirani and co show that the black box tests can be just as easily explained with another classical algorithm as by quantum annealing. “We outline a simple new classical model, and show that on the same data it yields correlations with the D-Wave input-output behavior that are at least as good as those of simulated quantum annealing,” they say.
That means that if the D-Wave computer wasn’t quantum at all, it would still be capable of producing the same results.
If correct, that’s a serious setback for D-Wave, which bills itself as “the quantum computer company”.
Vazirani and co’s paper is already sending ripples through the quantum computing community, which has been largely hostile towards D-Wave’s claims for its device. “Was previous skepticism too GENEROUS to D-Wave?” asks Scott Aaronson, a theoretical computer scientist at the Massachusetts Institute of Technology and self-appointed (twice retired) Chief D-Wave Skeptic, in his blog which has provided ongoing commentary on the debate.
Of course, D-Wave can still argue that its machine still is quantum but in a way that is not revealed in these tests. But at some point it’ll need to produce evidence to back up this claim.
What this entire episode exposes is the tension between D-Wave’s claim that it is selling a commercial quantum computer and many other physicists’ belief that the device is just a quantum experiment, the outcome of which has yet to be properly determined. Most physicists fully expect a useful quantum computer to eventually emerge, just not in the way D-Wave proposes.
At the moment, the weight of evidence is tipping dangerously against D-Wave, leaving its customers in a precarious position. Could it be that Google, NASA and Lockheed Martin have shelled out tens of millions for a cryogenically-cooled calculator of the classical variety? According to the IBM and UC Berkeley team: just possibly, yes.
Ref:arxiv.org/abs/1401.7087 : How “Quantum” is the D-Wave Machine?