Chinese team’s classical computing tackles the ‘impossible’ to challenge Google’s ‘quantum supremacy’
- Chinese Academy of Sciences team says it has developed an algorithm to perform a task ‘thought to be impossible for classical computations’
- Researchers say the 1 million uncorrelated samples from their method have a greater fidelity than that of the Google quantum computer
The Chinese team said their non-quantum classical computer completed the sampling task “in about 15 hours” with higher estimated fidelity – or accuracy – than Google’s quantum computer Sycamore, which took 200 seconds for the same task.
The team said the 1 million uncorrelated samples generated using their method had a fidelity of 0.0037, compared to that of the Google quantum computer’s 0.002.
In a paper to be submitted to a scientific journal for peer review, scientists at the Institute of Theoretical Physics under the Chinese Academy of Sciences said their algorithm on classical computers completed the simulation for the Sycamore quantum circuits “in about 15 hours using 512 graphics processing units (GPUs)”.
“We propose a new method to classically solve this problem by contracting the corresponding tensor network just once, [it] is massively more efficient than existing methods in obtaining a large number of uncorrelated samples with a target fidelity,” they said.
That claim – particularly how Google scientists arrived at the “10,000 years” conclusion – has been questioned by some researchers.
In Beijing, the team at the Chinese Academy of Sciences argued that “the computational time estimated by Google relies on a specific classical algorithm … rather than a theoretical bound that applies to all possible algorithms”.
“So, in principle, there could exist algorithms that perform much better than the algorithm used by Google, rejecting the quantum supremacy claim,” the team said. “We provide such an algorithm based on the tensor network method.”
Team leader Zhang Pan, a professor at the institute, told the South China Morning Post that currently in quantum computing it was important to combine classical and quantum computing – which suffers from “noise”, or interferences that lower its accuracy – for real-world applications.
Unlike classical computing, quantum computing is prone to errors because subatomic behaviour can be affected by environmental factors.
“Our new algorithm and the use of advanced classical computational resources, including more than 500 GPUs, are why our device is comparable to Google’s quantum computer on this random quantum-circuit sampling problem,” he said.
“To further speed up a task and increase the fidelity, an even better algorithm could be developed and applied on equipment of larger scale and higher performance level, like a classical supercomputer.”
In their latest paper released this month, the team said “if our simulation of the quantum supremacy circuits can be implemented in a modern supercomputer with high efficiency, in principle, the overall simulation time can be reduced to a few dozens of seconds, which is faster than Google’s hardware experiments”.
“To the best of our knowledge this is the first time that the sampling problem of the Sycamore quantum supremacy circuits with fidelity larger than Google’s hardware samples … is solved in practice classically.”
Simulating quantum computation on traditional computers is important for developing the technology – it allows researchers to test and verify experiments before running them on a quantum machine, and can add to the understanding of which features are powering that machine and where the boundary lies between quantum and classical systems.
Quantum computing is still in its infancy but promises to take computational power to a new level by manipulating subatomic particles. Scientists hope it will help lead to breakthroughs in areas such as materials science and developing new drugs.