arXiv:2610.26784
High-Accuracy Propagation-Assisted Local Decoding for Fast Transversal Clifford Circuits
Mark Webber, Simanraj Sadana, Matthew A. Lane, Lucy Jade Robson, Anthony O'Rourke
Abstract
We benchmark the Iterative Redecode Scheme (IRS) on fast transversal Clifford circuits, achieving high-accuracy correlated decoding using local memory-style detectors and decoders. On a deep random Clifford benchmark, IRS yields approximately $4\times$ lower logical failure rates than the ghost protocol at $d=11$ with matched buffers, and $15\times$ lower rates than logical observable matching (LOM). In three-block GHZ preparation, IRS achieves $2\times$ lower logical error rates than reliable logical Pauli-product decoding with matched initialisation, improving to $25\times$ at $d=19$ with $\lceil d/2\rceil$ full initialisation rounds. We place IRS, which builds on Wan et al.'s scheme, within a broader propagation-assisted local (PAL) framework, where local decoders exchange messages rather than decoding a global hypergraph. IRS alternates local minimum-weight perfect matching (MWPM) on memory-equivalent detector error models (DEMs) with Pauli-frame propagation and detector updates across transversal CNOT layers, seeking local error estimates consistent with the global detector record. Prior-informed DEMs account for probabilities of propagated errors missing from the memory-equivalent model, improving initial decoding and reducing logical error rates. Zero-weight physical Pauli mechanisms enable inference and propagation of initially indeterminate stabilizer outcomes. Error rates saturate after a few iterations even for deep, adversarial alternating-tCNOT circuits, allowing iteration caps with negligible loss of accuracy. We find syndrome-extraction buffers of order $\lfloor d/4\rfloor$ sufficient for distance-dependent error suppression in the demanding regimes studied. The approach extends to other local decoders and CSS codes; for repeated logical $S$ gates, IRS achieves approximately $4$ to $10\times$ lower logical error rates than LOM, depending on initialisation.
AI Summary
Fast transversal Clifford circuits create cross-block error correlations that ordinary memory decoders do not capture. This paper studies whether iterative exchanges between local decoders can recover those correlations without decoding a global hypergraph. The Iterative Redecode Scheme (IRS) develops an existing propagate-and-redecode procedure: local MWPM solutions infer Pauli frames at gate boundaries, propagated contributions are removed from working detector records, and decoding repeats toward self-consistency. Additions include propagation-informed local error-model priors during the first iteration and zero-weight mechanisms representing initially indeterminate stabilizers. The method is also applied to a rotated-surface-code logical $S$ construction. Circuit-level Monte Carlo benchmarks give approximately $4\times$ lower logical failure rates than ghost decoding and $15\times$ lower than logical observable matching (LOM) at $d=11$ on a fixed deep random Clifford circuit with matched syndrome buffers. For three-block GHZ preparation, correlated IRS halves the published reliable-logical-Pauli-product error rate at $d=19$ with matched initialization; using $\lceil d/2\rceil$ full initialization rounds increases the reduction to $25\times$, with additional preparation overhead. Repeated-$S$ benchmarks show roughly $4$–$10\times$ reductions relative to LOM. Adversarial alternating-CNOT tests show performance saturating after a few iterations; roughly $0.1\%$ of shots cycle. Buffers of order $\lfloor d/4\rfloor$ empirically suffice for distance-dependent suppression in demanding cases. These are finite-size numerical findings, not asymptotic guarantees. Convergence establishes consistency, not maximum likelihood. Runtime and memory scaling remain unquantified, and a validated windowed implementation for real-time decoding is left to future work.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 22 pages, 13 figures (including appendix). Supporting simulation circuits and numerical data available at https://doi.org/10.5281/zenodo.23475403