Detector error models and coherent errors

October 27, 2025

A schematic of how coherent noise interacts with the surface code showing the errors, the measured probability of events, and then the detector error model — A schematic of how coherent noise interacts with the surface code showing first the errors on the code, then the measured probability of events, and finally the derived detector error model

Coherent noise can arise from miscalibration of gates, crosstalk or state preparation errors and is often more detrimental to the logical performance of QEC experiments than Pauli noise. Decoders of QEC experiments require detector error models (DEMs) that describe the error mechanisms and their rates, according to a noise-informed model. We propose a noise-estimation method that can learn coherent or Pauli noise using only the syndrome information of QEC experiments. When coherent noise is present, we find that the errors can interfere, leading to enhanced physical error rates. Additionally, we find that hyperedges can appear in the DEMs, which is a feature not observed for the Pauli-twirled models. Our noise estimation method is capable of informing the decoder of such features which lowers the logical error rate. We showcase this behavior through repetition and surface code simulations in code capacity, phenomenological and circuit-level noise models. Our code to simulate circuit-level coherent errors for a repetition code is available in https://github.com/eva-takou/Estimating_coherent_errors_with_DEMs. The results are posted on the arXiv: https://arxiv.org/abs/2510.23797.