Better accounting for qubit errors boosts surface code

Manuel Rispler of Forschungszentrum Jülich and colleagues have achieved a 6 percent error threshold for the surface code under uniform noise, a substantial improvement over previous methods. The team mapped the decoding problem to a disordered spin model, the random coupled-plaquette gauge model (RCPGM), which accounts for Y-errors, a major bottleneck in surface code decoders, in noisy measurements. This work also demonstrates an improvement to 1.4 percent when applied to more complex circuit-level noise, compared to 0.7 percent using an earlier model, expanding the possibilities for statistical mechanical mappings in quantum error correction. Researchers at Forschungszentrum Jülich and collaborating institutions have demonstrated this improvement over previous methods by accurately accounting for challenging Y-errors, a persistent bottleneck in building…