Title:The surface code under generic $X$-error channels: Statistical mechanics, error thresholds, and errorfield double phenomenology

Abstract:We study the surface code under the most general single-qubit $X$-error channel, encompassing both coherent and incoherent errors. We develop a statistical mechanics mapping for the decoding problem and represent the partition function as a transfer-matrix, i.e., a (1+1)$D$ hybrid quantum circuit. The relevant circuit evolution, as we show, has an efficient matrix product state approximation, using which we develop an algorithm for large-scale syndrome sampling, thus enabling the simulation of these non-Pauli errors away from the limits captured by previous methods. Using these results, we compute maximum-likelihood thresholds and show that error coherence has negligible influence away from the fully coherent limit. We also study how coherence impacts information-theoretic measures like the coherent information, which we show to require, for accurate threshold estimates, increasingly large codes upon increasing error coherence and to eventually break down in the fully coherent limit. To interpret the dominant effect of incoherent errors in large codes, we develop a phenomenological errorfield double field theory, which indicates the instability of coherent errors' above-threshold quasi-long-range order to the above-threshold disorder characteristic of bit-flip errors and the suppression of coherence in the logical noise channel.