Google claimed to have achieved a desired level of quantum computing excellence with its Sycamore system in 2019. But the problem it used to make the claim has just been solved by today’s accelerator of choice, the GPU. As reported by Science, researchers in China recently managed to solve the same computational problem that led Google to claim the title, despite being equipped with “only” 512 GPUs, which led to some clever changes to the original algorithm. However, the whole concept of quantum supremacy refers to the moment when a quantum computer solves a problem that would be impossible for a classical computer.
At the time, Google said it would take the fastest supercomputer at the time – provided by IBM Summit – an ungodly 10,000 years to solve the same calculation that its Sycamore quantum computer did in 200 seconds. The Chinese team’s 512 GPUs took fifteen hours to do the same.
It’s just another reminder that both time and quantum computing are relative—which is understandable, given the relative nascent state of the technology.
Google’s claim to the title of quantum supremacy was based on the discovery of an interference pattern in the qubit values. Because quantum computing is an erratic master, all current approaches to it are prone to decoherence, which refers to how the environment and design and operation of the qubit introduce errors into its calculations.
From these operational errors, and by running the same algorithm through Sycamore for 200 seconds (and millions of iterations), Google then extrapolates a result showing the processor’s pattern of deviations from the exact, correct values it should be outputting. These biases occurred because errors made certain outcomes more likely than others; this pattern was eventually visualized by a spike plot that could be reliably reproduced.
This graphical representation of the relationship between errors and results is what Google claims gives it quantum superiority. And this same graph was achieved by Chinese scientists. To achieve this, they represented the problem in terms of a 3D mathematical array – a matrix – which allowed the specialized tensor cores on their 512 GPUs to solve it by simply multiplying the values in the array.
“I think they’re right that if they had access to a big enough supercomputer, they could simulate … the task in seconds,“ Scott Aaronson, a computer scientist at the University of Texas, said Science. The Chinese team puts this estimate at 12 seconds of computing time.
To be fair, the Google scientists left a caveat in their article. Sergio Boixo, chief scientist for Google Quantum AI, said in an email to Science that “classical algorithms would improve”. And they improved it—perhaps a little too quickly, blunting the edge of Google’s claims and ultimately proving IBM’s objections correct.
But the Google engineers emphasized one point: technology is forever evolving, and quantum computing is now going through a stage of leaps and bounds that is now rare for classical systems. If Google’s Sycamore was able to provide sharp outlines with greater precision than that (at 0.2%), today’s quantum computers would do better, due to improvements in error correction.
The low accuracy achieved by Sycamore was exactly the part that gave the Chinese scientists some leeway – they simply improved the accuracy of their calculations to 0.37%. That’s enough to beat Sycamore, but it’s still far from what’s theoretically possible. That fact, and the nature of quantum computers under development, led Sergio to add that “we don’t think this classical approach can handle quantum circuits in 2022 and beyond.”
And while that’s also very likely correct, it looks like Google should take the Quantum Supremacy trophy off its wall. Other hands will surely go up to claim it. It’s only a matter of probability – and as such it’s also only a matter of time.