Large-Scale Open-System Simulation for Hardware-Aware QEC Decoding

Decoder performance is a major bottleneck for quantum error correction (QEC). We address this by constructing a hardware-aware decoder directly from large-scale, pulse-level open-system simulations that incorporate realistic control, cross-talk, syndrome extraction, and noisy readout over many cycles. A real-time quantum Monte Carlo (QMC) engine provides the computational scale to study distance-3 and distance-5 surface codes without compromising physical fidelity. From these simulations we perform efficient process tomography on simulated data to estimate process characteristics and translate them into simulation-derived edge weights for minimum-weight perfect matching (MWPM). The learned weights faithfully encode time-dependent, hardware-realistic noise processes that phenomenological, hand-tuned empirical fits fail to capture. Over the regimes studied, the simulation-derived MWPM decoder achieves lower logical error rates and yields more reliable threshold projections . This simulation-first workflow—accurate open-system modeling, process estimation on simulated datasets, and principled weight construction—provides a practical route to hardware-aware decoding that delivers experiment-ready gains in QEC.