Google Claims Error Correction Will Enable Quantum Computers to Be Useful

This can mean that adding physical qubits in your logical qubit could actually prove to be harmful. “So far engineers have tried to organize larger and more complex ensembles of physical qubits in logical qubits in order to achieve lower error rates. The opposite happened,” states Hartmut at Google.

This was demonstrated by Google’s announcement of a working error-correction scheme in 2021. It resulted in an increase in errors. The Joint Quantum Institute in Maryland achieved a level where logical qubits did not worsen error rates. However, this was at a technical rather that practical level.

Google now shows that logical qubits are possible to be made larger and that this increases the error rate. If this trend continues, quantum computers will be able to perform computations that are impossible on any of the most powerful classical computers. Neven stated that there is “palpable confidence” in the team that Google will make a commercially viable quantum computer.

The third generation of Google’s Sycamore, which has 53 qubits, was used by the team to achieve its logical qubit milestone. The Surface code logical quibits are a grid of qubits that have been paired with one of the same size. One qubit is reserved for measuring the value of other qubits. The experiment by the company saw a shift from 3 by 3, which involved 17 physical qubits to 5 by 5, which used 49 qubits. This meant that almost all of the processor was acting as one logical qubit. This increased error rate saw a decrease from 3.028 to 2.914%.

Google admits that there is a small improvement, but the team says that it can be continued indefinitely. This opens the door to a fault-tolerant quantum computing system that can reliably perform useful tasks. However, moving to a 6×6 logical qubit (which would require 71 physical qubits) is not possible with Google’s current quantum processors. This will require a significant hardware upgrade.

Fernando Gonzalez Zalba from the University of Cambridge said that while it would have been nice to see an improvement in error rates, the research is moving in a positive direction.

He says that “the individual components of the processor need improvement to get an increase in the logical errors rate as technology scales.” “But, what we see in the series publications that the team produces is that they are significantly improving after each publication. While I doubt they will see quantum error correction in scalable scale for many years, I believe they are very close.