Information-efficient decoding of surface codes
Surface codes are a popular error-correction route to fault-tolerant quantum computation. The so-called exponential backlog problem that can arise when one has to do logical $T$-gates within the surface code demands real-time decoding of the syndrome information to diagnose the appropriate Pauli frame in which to do the gate. This in turn puts a minimum requirement on the communication rate between the quantum processing unit, where the syndrome information is collected, and the classical processor, where the decoding algorithm is run. This minimum communication rate can be difficult to achieve while preserving the quality of the quantum processor. Here, we present two decoders that make use of a reduced syndrome information volume, relying on a number of syndrome bits that scale only as the width -- and not the usual area -- of the surface-code patch. This eases the communication requirements necessary for real-time decoding.