Google’s quantum computer has reached a milestone in error correction


To reduce the error rate in quantum computers, sometimes more is better than more. That is, more qubits.

The quantum bits, or qubits, that make up a quantum computer are prone to errors that can render calculations useless if left uncorrected. To reduce this level of errors, scientists strive to create a computer that can correct its own errors. Such a machine would combine the capabilities of several false qubits into one advanced qubit, called a “logical qubit,” that can be used for computation.

Now scientists have demonstrated a key milestone in quantum error correction. Increasing the number of qubits in a logical qubit could make it less error-prone, Google researchers report February 22 in Nature .

Quantum computers of the future can solve problems that are impossible even for the most powerful traditional computers. To build these powerful quantum machines, researchers agree that they will need to use error correction to dramatically reduce the error rate. Although scientists have previously demonstrated that they can detect and correct simple errors in small quantum computers, error correction is still in its early stages.

The new progress doesn’t mean researchers are ready to build a fully error-corrected quantum computer, “but it does demonstrate that it is indeed possible, that error correction works in principle,” physicist Julian Kelly of Google Quantum AI said in a news release. briefing on February 21.

Quantum computers like Google’s require a dilution refrigerator (pictured) that can cool the quantum processor (mounted at the bottom of the refrigerator) to low temperatures.GOOGLE QUANTUM AI

Logical qubits redundantly store information in multiple physical qubits. This redundancy allows a quantum computer to check for errors and correct them on the fly. Ideally, the larger the logical qubit, the lower the error rate should be. But if the original qubits are too faulty, adding more will cause more problems than they solve.

Using Google’s Sycamore quantum chip, the researchers studied two different sizes of logical qubits, one with 17 qubits and the other with 49 qubits. After steadily improving the performance of the original physical qubits that make up the device, the researchers counted the errors that still occurred. The larger logic qubit had a lower error rate, about 2.9 percent per round of error correction, compared to about 3.0 percent for the smaller logic qubit, the researchers found.

This small improvement suggests that scientists have finally tiptoed into a regime where error correction can begin to suppress errors by scaling up. “This is the main goal to be achieved,” says physicist Andreas Wallraff of ETH Zurich, who was not involved in the study.

However, the result is just on the verge of demonstrating that error correction improves as scientists scale up. Computer simulations of the performance of a quantum computer suggest that if the size of a logical qubit is increased even more, its error rate will actually become worse. Further refinement of the original faulty qubits will be required to enable scientists to truly benefit from error correction.

Still, milestones in quantum computing are so difficult to achieve that they’re seen as leaps and bounds, Wallraff says. You just want to barely break the bar.

Share post:



More like this