Google says fixing bugs will lead to useful quantum computers

Google has shown for the first time that it is possible to reduce the total number of errors generated by a quantum computer, meaning that it will be possible to build larger and more useful devices.

Google has demonstrated that its approach to quantum error correction, which is seen as an important part of developing useful quantum computers, is scalable, giving the company’s researchers confidence that practical devices will be ready in the coming years.

The building blocks of a quantum computer are qubits, similar to the transistors in a classical computer chip. But today’s qubits are subject to interference and errors that need to be identified and corrected if we are to build quantum computers big enough to solve real-world problems.

One popular approach to this is called surface code patching, in which many physical qubits operate as one so-called logical qubit, essentially introducing redundancy. This is how much error correction works in classical computers, but there are additional complications in quantum computers because each qubit exists in a mixed superposition of 0s and 1s, and any attempt to measure them directly destroys the data.

This means that adding more physical qubits to your logical qubit can be detrimental. “Until now, when engineers tried to organize larger and larger ensembles of physical qubits into logical qubits to reduce the error rate, the opposite happened,” says Google’s Hartmut Neven.

Google demonstrated this when it first announced a working bug fix scheme in 2021, resulting in a net increase in bugs. Subsequent work at the Joint Quantum Institute in Maryland reached a point where logic qubits did not worsen the error rate, albeit on a technical rather than a practical level.

Google has now shown that the size of logical qubits can be increased, and that such scaling leads to a reduction in the overall error rate. If this trend can be continued and the scale of quantum computers can be increased, then they will be able to perform calculations that would be impossible even on the most powerful classical computers. Neven says there is now “palpable confidence” among the team that Google will build a commercially useful quantum computer.

The team reached their qubit logic milestone using Google’s third-generation Sycamore quantum processor, which has 53 qubits. The logical qubits of a surface code are usually a grid of qubits connected to another qubit of the same size, with one qubit reserved to measure the value of the others. The company’s experiment went from a 3-by-3 grid involving 17 physical qubits to a 5-by-5 ​​grid using 49 qubits, meaning almost the entire processor acted as a single logical qubit. This increase resulted in a reduction in the error rate from 3.028% to 2.914%.

The Google team acknowledges that the improvement is small, but says the scaling process could theoretically continue indefinitely and paves the way for a fault-tolerant quantum computer that can reliably perform useful tasks. But moving to a 6-by-6 logical qubit, which would include 71 physical qubits, is not possible with the company’s current generation of quantum processors and would require a big leap forward in hardware.

Fernando González-Zalba of the University of Cambridge says it would be nice to see more reduction in error rates, but research is moving in the right direction.

“Individual processor components need to be improved a bit to achieve lower logic error rates as the technology scales,” he says. “[But] what we see in the series of publications that the team puts out is that they improve substantially after each publication. I don’t think we’re talking years before we can see scalable quantum error correction, I think they’re pretty close.”