Ů��С��Ƶ

After a month of waiting, Google has finally broken its silence on quantum supremacy. On 23 October, engineers at the firm released a paper claiming that they have built a quantum computer that can solve a problem that cannot realistically be solved by a traditional computer – the accepted definition of quantum supremacy.

An earlier version of the paper was accidentally posted to a NASA server in September, but Google has refused to discuss its achievement until now. In the intervening time, researchers at IBM examined Google’s claim of quantum supremacy and found it lacking.

They found that, while Google claimed it would take IBM’s Summit supercomputer 10,000 years to solve the random number problem that the quantum computer solved, it could actually take the supercomputer less than three days. It took Google’s Sycamore quantum computer 3 minutes and 20 seconds. But if the supercomputer could perform the same task over a long weekend, that would mean the random number problem is not realistically beyond the reach of traditional computers. In other words, Google would not actually have achieved quantum supremacy.

Test it

In a press call, two of the Google researchers involved in the work hit back at IBM’s claims. “We’re looking forward to when people actually run the idea on Summit and check it and check our data because that’s part of the scientific process – not just proposing it but actually running it and checking it,” said John Martinis. “At the same time, we’ll be making our quantum computers better.” We can’t be sure that IBM’s algorithm actually works until it is properly tested, he said.

“Even with the two-day possibility they have to run this, it still takes longer than the 200 seconds we say it should take, so we’re still in the quantum supremacy regime,” said Martinis. “We expect that quantum computers will [eventually] vastly outstrip what’s going on with the algorithms.”

These tests of quantum supremacy claims are good for the field, said Sergio Boixo. “Now that the paper is out showing the data and the circuits – and we put it out specifically for this reason – we encourage other groups to analyse the data.”

Algorithms on classical computers will no doubt continue to improve and possibly outstrip this particular quantum computer, but quantum computers could improve far faster, said Boixo.

The researchers also discussed the future of their device, including eventually making it available to the public, and quantum computing more generally. “We always like to say that quantum computing is a very long-term research project, and it’s going to take years more to get to the point where we’re doing more useful work,” said Martinis. “The Sycamore device is kind of the first in a series of devices that we plan to make. We plan to make them better and scale them up.”

While this device had 53 working quantum bits, or qubits, future devices will need more to perform simulations of actual quantum processes. “One of the experiments that we want to do is around 1000 qubits in the next few years,” said Martinis. “There are clearly a lot of hard things to solve, but we are very excited about that.”