Abstract We study decoder-representation sensitivity in simulated rotated surface-code Z-memory circuits with local spatially correlated XX pair events. At distance 3, ordinary minimum-weight perfect matching (MWPM) produced 4,610 versus 1,260 logical failures at p_corr = 10^-4 and 5,437 versus 2,239 at p_corr = 2 × 10^-4 on the same 500,000-shot samples using original-order versus canonical-sorted detector error models (DEMs). Across 50 additional seeded permutations per condition, failure counts varied widely. Canonical sorting tied for the minimum in this finite tested set; it is not claimed to be universally optimal. The ordering experiment reorders complete error instructions in a flattened DEM, preserving their probabilities and absolute detector and logical targets. PyMatching documents retaining the first-added logical-label set when parallel graph edges are merged; we therefore interpret the findings as a workflow-specific reproducibility result, not a newly discovered general software defect. At distances 5 and 7, correlation-aware MWPM had lower observed failure rates in all eight tested conditions, with six paired contrasts remaining significant after Bonferroni correction. An exact maximum-a-posteriori calculation gave a 0.2367% Bayes risk for the flattened-DEM distribution at p_corr = 10^-4; this is optimal for that modeled distribution only. This bounded classical simulation study proposes no new decoder, estimates no threshold, and provides no hardware evidence. Some historical source-to-result provenance remains incomplete, as disclosed in the manuscript. Code and evidence: https://github.com/tgiwn/QEC-surface-code-correlated-noise/tree/106d32e815c5f6009941afc36d3f65be7ecf35a3 Status: This is a preprint and has not undergone peer review.
Abstract We study decoder-representation sensitivity in simulated rotated surface-code Z-memory circuits with local spatially correlated XX pair events. At distance 3, ordinary minimum-weight perfect matching (MWPM) produced 4,610 versus 1,260 logical failures at p_corr = 10^-4 and 5,437 versus 2,239 at p_corr = 2 × 10^-4 on the same 500,000-shot samples using original-order versus canonical-sorted detector error models (DEMs). Across 50 additional seeded permutations per condition, failure counts varied widely. Canonical sorting tied for the minimum in this finite tested set; it is not claimed to be universally optimal. The ordering experiment reorders complete error instructions in a flattened DEM, preserving their probabilities and absolute detector and logical targets. PyMatching documents retaining the first-added logical-label set when parallel graph edges are merged; we therefore interpret the findings as a workflow-specific reproducibility result, not a newly discovered general software defect. At distances 5 and 7, correlation-aware MWPM had lower observed failure rates in all eight tested conditions, with six paired contrasts remaining significant after Bonferroni correction. An exact maximum-a-posteriori calculation gave a 0.2367% Bayes risk for the flattened-DEM distribution at p_corr = 10^-4; this is optimal for that modeled distribution only. This bounded classical simulation study proposes no new decoder, estimates no threshold, and provides no hardware evidence. Some historical source-to-result provenance remains incomplete, as disclosed in the manuscript. Code and evidence: https://github.com/tgiwn/QEC-surface-code-correlated-noise/tree/106d32e815c5f6009941afc36d3f65be7ecf35a3 Status: This is a preprint and has not undergone peer review.
Abstract Quantum error-correction decoders must convert noisy syndrome records into logical decisions while operating under strict latency and reliability constraints. Minimum-weight perfect matching (MWPM) remains a strong baseline for surface-code decoding because it is fast, interpretable, and exact for its graphlike optimization problem [1–8]. However, the graphlike independent-error approximation can become mismatched when syndrome records contain temporal drift, measurement corruption, crosstalk, bursts, or mixtures of mechanisms [9–18]. We introduce SAGE-QEC, a failure-aware selective decoder that retains nominal PyMatching as its default decision and learns when an observable syndrome structure warrants an alternative candidate. SAGE-QEC uses train-frozen structure thresholds, a candidate ranker, validation-only policy selection, and a paired blind evaluation. Across 45 simulator conditions, SAGE-QEC reduced the aggregate logical error rate from 27.1285% for PyMatching to 24.8320%, corresponding to an 8.47% relative reduction. The campaign used rotated surface-code memory circuits at distances 3, 5, and 7, nominal circuit parameters 0.002, 0.003, and 0.005, five seeds, and four test regimes withheld from training: Burst-3, Temporal Drift, Crosstalk, and Mixed Noise. The test set contained 81,000 paired shots. All twelve distance-by-regime aggregate comparisons favored SAGE-QEC in the executed campaign, with paired bootstrap intervals below zero for SAGE-QEC minus PyMatching. The result is evidence of a reproducible simulator-level improvement, not a claim of universal decoder superiority. SAGE-QEC currently uses logical-output proxy candidates rather than a complete physical correction search, and its intervention rate remains high under distribution shift. It has not yet been compared with a fully calibrated correlated-noise space-time PyMatching baseline or hardware calibration data. We therefore characterize SAGE-QEC as a strong qualification-stage prototype and release a complete audit protocol for the next industrial validation stage.
We describe a decoder for the rotated surface code that runs as parallel tempering on probabilistic bits (p-bits) built from single-flux-quantum (SFQ) cells at 4.2 K. The X graph is decoded first, and its result switches the weights of the Z graph to account for X–Z correlations from Y errors. The p-bits, the edge toggle cells with merger trees for up to five writers and the timing of the colour clock are simulated at circuit level in JoSIM, and the measured properties enter a GPU model of the full decoder. Over d rounds of circuit-level noise at p = 0.006, paired shot by shot, the decoder has 14 to 18 % fewer logical failures than minimum-weight perfect matching (MWPM) at d = 5 and 27 to 36 % fewer at d = 7 (two independent seeds) and, with both seeds pooled, 2.2 % (d = 5) and 4.7 % (d = 7) more failures than correlated matching. Without the weight switching its logical outcome agrees with MWPM in almost every shot, so the gain comes from the X–Z correlation step (97 to 104 % of it); a deterministic two-stage control with the same switching rule comes within 2 % of correlated matching, so most of the remaining gap is due to the sampler. At p = 0.004 and 0.008 the decoder has 15 to 48 % fewer failures than MWPM (one seed). A block of d rounds is decoded in 0.53 µs (d = 5) and 1.67 µs (d = 7). Counted from the simulated cells, the core of the d = 7 decoder (p-bits, toggle cells, splitters, mergers; control logic and wiring excluded) needs 6.82 million Josephson junctions and 11 to 22 mW under energy-efficient RSFQ (ERSFQ) biasing. The decoder is a simulation study; it has not been built. Files. P5_manuscript.pdf: the manuscript. sfq_pbit_decoder_p5.tar.gz: simulation and analysis scripts (JoSIM test benches, GPU decoder model and its patches, evaluation scripts), JoSIM raw results, per-shot logical outcomes of all decoder runs, the receipts from which every number and figure is generated, the automated number checks and a self-test that reruns all evaluations and compares them field by field with the receipts. See README.md in the archive. Licences. Text and data CC BY 4.0, code MIT. Use of AI. Large language models (Claude, Anthropic) wrote most of the simulation and analysis code, ran the simulations, searched and read the literature, produced the figures and drafted the text under the author's direction; AI agents were also used as adversarial reviewers. Every number is generated by scripts from the raw simulation output and checked automatically. Version 1.1 (4 November 2026): references to work on SFQ random number generators for stochastic computing and Bayesian networks (Yamanaka, Yoshikawa and Yamanashi, IEEE Access 2024; Yamanashi, ISS 2025) added. Results, numbers, figures, code and data are unchanged.
Accurate decoder priors are essential for high-performance quantum error correction. Theoretical noise models inevitably omit physically relevant error mechanisms, most notably correlated errors, which appear consistently in experimental error budgets yet are absent from standard models. We introduce a data-driven method for augmenting detector error models (DEMs) by discovering entirely new error mechanisms directly from experimental detector-firing statistics, going beyond most existing approaches that merely refine the probabilities of already-known mechanisms. Our method generates candidates by combining pairs of existing mechanisms as building blocks, then filters and validates them using pairwise and higher-order detector correlations. Crucially, this construction provides each newly discovered mechanism with a well-defined logical effect derived from the constituent error mechanisms, enabling its direct use in decoding. Furthermore, each mechanism is accompanied by a natural decomposition into lower-order components, preserving compatibility with the original decoder. This is particularly valuable in settings that rely on correlated graph-based decoders such as correlated matching. We evaluate the method on surface-code simulations with correlated noise and correlated-matching decoding, demonstrating that, in a realistic near-term noise scenario, DEM augmentation increases the Lambda parameter by 20% across code distances 7-15 and reduces the logical error rate by 75% at code distance 15 relative to the unaugmented DEM. On experimental data from Google's superconducting device, the augmented DEM increases the Lambda parameter by 9% across code distances 5-7 and yields a reduction of 14% in logical error rate over the baseline for distance 7. To our knowledge, this is the first data-driven DEM augmentation method that produces decomposed hyperedges with assigned logical effects.
Efficient and realistic error decoding is crucial for fault-tolerant quantum computation (FTQC) on near-term devices. While decoding is a classical post-processing task, its effectiveness depends on accurately modeling quantum noise, which is hardware-dependent. In particular, correlated bit-flip (X) and phase-flip (Z) errors often arise under circuit-level noise. We introduce the Iterative Reweighting Minimum-Weight Perfect Matching (IRMWPM) decoder, which systematically incorporates such correlations to enhance quantum error correction. Our method leverages fault-detection patterns to guide reweighting: correlated X and Z detection events are identified, and their conditional probabilities update weights on the primal and dual lattices. This iterative procedure improves handling of realistic error propagation in a hardware-agnostic yet noise-aware manner. We prove that IRMWPM converges in finite time while preserving the distance guarantee of MWPM. Numerical results under circuit-level noise show substantial improvements. For distances ≥17 and physical error rates ≤0.001, IRMWPM reduces logical error rates by over 20x with only a few iterations. It also raises the accuracy threshold from 1% to 1.16%, making it practical for near-term real-time decoding. Extrapolated estimates suggest that to reach logical error rate 10−16, IRMWPM requires distance d=31, while standard MWPM needs d=50, implying a major reduction in qubit overhead.
Abstract Spatially concentrated noise is a central concern in modular quantum error correction, yet its effect is often estimated by replacing a heterogeneous device with a uniform effective error rate. We show that matching the expected number of physical faults does not preserve the logical failure probability in rotated surface-code memories. First, we reanalyse a published 2.6-million-shot proxy benchmark and identify six specific implementation issues that constrain its interpretation: an undocumented seam fraction of 0.5 instead of the configured 0.3, noise injection outside the repeated circuit body, an X error appended after measurement that cannot flip the recorded outcome, an unused p_ebit field, an algebraically generated decay curve presented as data, and a resource annotation not matching any circuit layout. Second, we perform 14.4 million additional sampled memory experiments across 72 conditions, covering code distances three to nine, both logical memory bases, spatially resolved seam noise, a channel-wise mean-matched control, and decoder mismatch. All numerical comparisons use recorded failure counts, with binomial confidence intervals and multiplicity-adjusted hypothesis tests. At code distance three, X-basis memory, and the largest tested noise severity, the geometric seam produces 977 failures while a channel-wise mean-matched homogeneous control produces 1430 failures in 200,000 shots, giving a risk ratio of 0.683 with a 95% confidence interval of 0.630 to 0.741 (Holm-adjusted p = 4.6 × 10⁻¹⁹). Only this d = 3 X-basis comparison is statistically decisive at the available shot count; the larger-distance observations are consistent with the same effect but are individually under-powered. Decoding the same geometric-noise samples with calibrated matching weights reduces failures from 2060 to 977 relative to baseline weights. Z-basis point estimates show the opposite geometry trend, although the selected largest-severity comparisons do not survive familywise adjustment. A simple repetition-code calculation explains why matching the mean physical error probability cannot generally preserve logical risk. The results establish a reproducible method for separating spatial exposure, temporal exposure, and decoder assumptions in seam-noise studies. They support geometry-aware modeling and calibrated decoding within the tested memory circuits; they do not implement a distributed gate, a Bell-pair channel, or classical feedforward, and they do not establish a distributed-gate threshold or hardware-level resource advantage.
We consider the rotated surface code of odd distance d with its standard syndrome-extraction circuit, under circuit-level depolarizing noise in which every gate, idle step, preparation and measurement fails independently with probability p, decoded by minimum-weight perfect matching with the log-likelihood weights of single faults. We prove that a memory experiment with T rounds fails with probability at most 0.061 (T+1)(d+1) λd/(1−λ), where λ = 0.999 (p/p**)9/25, for every p ≤ p** = 2×10−3. Hence the logical error rate per round decays exponentially in d for all p ≤ p**. Earlier rigorous thresholds for matching on surface-code circuits are 1.7×10−4, due to Dennis, Kitaev, Landahl and Preskill, and 7.4×10−4, due to Fowler. We keep their reduction of a failure to a weighted self-avoiding path in the decoding graph, but bound these paths by a local criterion for non-backtracking walks, which gives the threshold 1.04×10−3 in closed form. To reach 2×10−3 we use the minimality of the correction along the path, for a canonical choice among the minimum-weight corrections: labellings of short stretches of the path into error and correction edges that contradict this minimality are excluded, and the remaining labelled walks are counted with a potential function that is checked in exact arithmetic. For a leaner schedule with fused measure-and-reset steps the two arguments give 1.53×10−3 and 3×10−3, respectively. With error rates that depend on the type of operation, they certify exponential decay for error models built from the published component error rates of Quantinuum's Helios and H2 and of IBM's Heron r3 processor ibm_boston, mapped onto this circuit. All theorems are also proved in the proof assistant Lean 4. Any argument that bounds the failure probability by the probability of an increasing event, ours included, is limited by the smallest such event that contains all failures, and Monte Carlo estimates place the threshold of this event near 4×10−3 for this circuit. Changes from version 1: a third exclusion rule, which uses a canonical choice among the minimum-weight corrections, raises the certified thresholds to 2×10−3 for the standard schedule and 3×10−3 for the lean one and strengthens the certificates of the device error models. Lean 4 proofs of all theorems are new in this version. Programs, certificates, formal proofs and the outputs of all checks: doi:10.5281/zenodo.23485205. Mathematics Subject Classification (2020): 81P73; 94B35, 05C38, 68V05.