Google claims to have demonstrated quantum supremacy once again

Contents

Quantum computers are surrounded by a glow of great promises: more computing power, higher speed, better than any supercomputer. However, this high performance has yet to be demonstrated. Now a research team from Google Research claims to have solved a task with its quantum processor Sycamore that exceeds the computing capacity of classic supercomputers. This capability is often referred to as quantum supremacy.

First claims of quantum supremacy

Back in 2019, Google researchers claimed to have demonstrated quantum supremacy with an earlier Sycamore model. The chip with 53 superconducting qubits was said to have solved a task in 200 seconds that would have taken a supercomputer 10,000 years. The news received a great deal of media attention, but disillusionment followed just a few days later: competitor IBM showed that a supercomputer could solve this task in just 2.5 days using the right methods.

The main problem with the work at the time was that Google designed the calculation task specifically to be solved faster by a quantum computer than by a classic supercomputer. There was no mathematical proof of the superiority of the quantum computer, nor was there any practical benefit. The issue was simply so special that no one had ever tried to solve it efficiently with a supercomputer before.

Google is not the only company whose alleged quantum advantage has been disproved. In 2023, IBM claimed to have demonstrated that it had solved a useful mathematical problem with its 127-qubit processor. Only a short time later, researchers showed that a classical computer is also capable of doing this.

The solved problem

The latest publication by the research team led by Alexis Morvan, Benjamin Villalonga, Xiao Mi and Salvatore Mandra from Google Research and the Quantum Artificial Intelligence Lab at NASA Ames Research Center follows on from the 2019 results. It was published in the journal Nature. Frank Wilhelm-Mauch, Head of the Institute for Quantum Computer Analytics at Forschungszentrum Jülich, says: "This paper takes up the technology of the 2019 work and makes many things much better."

The task that the quantum computer is supposed to solve is called "Random Circuit Sampling". It is a quantum algorithm that generates a series of random numbers. The researchers execute a series of randomly selected quantum gates (i.e. quantum physical arithmetic operations) to bring the qubits into a complicated configuration that is difficult for classical computers to simulate.

"I am not aware of any practical use for random circuit sampling," says Sabine Wölk from the DLR Institute of Quantum Technologies. However, there are so-called variational quantum circuits, which could have a similar structure to the "random circuits" generated by Google. [It is hoped that these variational quantum circuits will provide a quantum advantage, for example in the calculation of chemical energies or in the field of machine learning."

Errors due to noise

Unlike in 2019, the Google researchers do not title their findings with the claim of quantum supremacy. In fact, the terms "quantum supremacy" or "quantum advantage" do not appear in the text. Instead, the publication is entitled "Phase transitions in random circuit sampling". Only in the abstract do the researchers write "that the computational costs of our experiment exceed the capabilities of existing classical supercomputers." In addition to demonstrating quantum superiority, it also addresses another research question.

Quantum systems are sensitive and even the smallest disturbances destroy the quantum information that is valuable for quantum computing. Effectively, this means that the results of a calculation become noisy and unusable after just a few operations. However, it is unclear how high the noise or error rate can be for a quantum computer to deliver usable results.

A low-noise phase

In their latest publication, the Google researchers now claim to have better understood the conditions under which a quantum computer could deliver an advantage over classical systems. In their experiment, the team increased the noise of the quantum processor in stages – so they deliberately disturbed the qubits to find out when the chip loses its quantum superiority.

They used a new version of the Sycamore processor with 67 superconducting qubits. They found that reducing the error rate meant that the results of the quantum simulation could no longer be simulated by a classical supercomputer. Furthermore, they estimate that it would take a supercomputer 10 trillion years to perform the same calculation.

Specifically, they observed two phase transitions. The latter refers to the transition of a system from one state to another in which it behaves significantly differently, as in the change from water to ice. In the low-noise phase, the calculations are so complex that quantum computers are superior to classical calculations.

For perfect quantum computers, it is easy to identify the threshold above which they are superior to classical computers, explains Wölk. This is not so easy for flawed quantum computers as they exist today. With the phase transition, the Google researchers were able to find such a threshold. "The results help to assess the quality of real quantum hardware and whether it is even possible to generate a quantum advantage with real existing quantum hardware." Overall, Wölk therefore considers the results of the current publication to be useful, in contrast to the results from 2019. "They help us to assess how far real quantum hardware has already come," she says.

Recently, quantum researchers have focused on reducing the error rate of their qubits and correcting errors that have occurred. Google's results now show that a quantum advantage could also be possible without error correction. "This cements the fact that quantum computers are superior to classical computers even if they are not perfect," says Wilhelm-Mauch. "However, it is still a synthetic benchmark that cannot easily be translated into a real application." It is therefore still unclear whether quantum superiority can also be demonstrated in other – more useful – problems.

(mma)