June 3, 2026
Quantum error correction (QEC) is essential for building large-scale, fault-tolerant quantum computers. However, platforms like trapped ions frequently experience biased noise, where dephasing errors occur significantly more often than bit-flip errors. While bias-tailored codes like Clifford-deformed elongated compass codes have shown promise in handling this asymmetry under simplified noise models, evaluating how they stack up under realistic, circuit-level simulations is a crucial hurdle for practical implementation.
In this work, we perform circuit-level simulations of Clifford-deformed elongated compass codes. We evaluate their performance by comparing standard Minimum Weight Perfect Matching (MWPM) against correlated MWPM decoding, which accounts for correlations created by hyperedges in the decoding graph. These faults are decomposed by standard MWPM. We find that correlated decoding consistently enhances thresholds across all noise biases. What is more, at the circuit level, Clifford deformations and code asymmetry amplify the performance of the correlated decoder. This demonstrates that the correlated decoder is suited to exploit the asymmetric structure of these codes, highlighting that matching the right decoder with code co-design is key to unlocking the maximal performance of hardware-tailored QEC.
The manuscript is now available on arXiv: https://arxiv.org/abs/2605.27598