Quantum error correction (QEC) decoders produce corrections but provide noper-shot signal of correction reliability. We introduce syndrome-consistency gating: adecoder-independent shot acceptance filter that uses redundant syndrome copies to identifyunreliable decoder outputs. The gate is veto-only (it may block shots but never overrides thedecoder) and fails closed (blocks all shots on gate failure). We validate this approach on twoindependent hardware vendors. The primary dataset comprises 768,000 shots on IBM Torino(Heron r2, 156 qubits); a cross-vendor dataset of 64,000 shots on Rigetti Ankaa-3 (84qubits) provides supporting evidence; and an exploratory dataset of 64,000 shots on IBMFez preceded both. Total: ~896,000 hardware shots. All experiments use repetition codes.On IBM Torino with a distance-5 code, strict gating with two redundant copies (f=2) achieves100% logical error rate (LER) reduction (0.188% to 0.00%) at 36.3% allow rate, with all 30observed logical errors concentrated exclusively in blocked shots. Under practicalparameters, LER reduction reaches 85.6% (0.081% to 0.012%) at 53.3% allow rate with13.7× error concentration. On Rigetti Ankaa-3, gated active broadcast achieves zero logicalerrors in 16,076 accepted shots versus 9 errors in 24,000 passive shots (p = 0.014). Wecharacterise the allow-rate scaling as exponential decay (R² ≥ 0.98) and identify f=2 as thepractical operating point.
Quantum error correction (QEC) decoders produce corrections but provide noper-shot signal of correction reliability. We introduce syndrome-consistency gating: adecoder-independent shot acceptance filter that uses redundant syndrome copies to identifyunreliable decoder outputs. The gate is veto-only (it may block shots but never overrides thedecoder) and fails closed (blocks all shots on gate failure). We validate this approach on twoindependent hardware vendors. The primary dataset comprises 768,000 shots on IBM Torino(Heron r2, 156 qubits); a cross-vendor dataset of 64,000 shots on Rigetti Ankaa-3 (84qubits) provides supporting evidence; and an exploratory dataset of 64,000 shots on IBMFez preceded both. Total: ~896,000 hardware shots. All experiments use repetition codes.On IBM Torino with a distance-5 code, strict gating with two redundant copies (f=2) achieves100% logical error rate (LER) reduction (0.188% to 0.00%) at 36.3% allow rate, with all 30observed logical errors concentrated exclusively in blocked shots. Under practicalparameters, LER reduction reaches 85.6% (0.081% to 0.012%) at 53.3% allow rate with13.7× error concentration. On Rigetti Ankaa-3, gated active broadcast achieves zero logicalerrors in 16,076 accepted shots versus 9 errors in 24,000 passive shots (p = 0.014). Wecharacterise the allow-rate scaling as exponential decay (R² ≥ 0.98) and identify f=2 as thepractical operating point.
Readout errors are one of the dominant sources of noise in current quantum processors, limiting both expectation-value estimation and sampling-based applications. Since they affect only the classical measurement outcomes, they can be addressed using classical coding techniques: immediately before measurement, each data qubit is redundantly encoded with ancilla qubits, and the resulting bit string is decoded either by post-selection or by majority voting. Unlike conventional readout error mitigation, which corrects only aggregate quantities such as expectation values, this approach operates on individual measurement shots and can therefore produce approximately corrected samples. We present a systematic cross-platform and cross-generation experimental evaluation of repetition-code readout error detection and correction. We benchmark the same protocol on IBM Heron r1-r3 superconducting processors and Quantinuum H1 and H2 trapped-ion processors while independently varying the code distance, hardware generation, and encoding layout. We find that both error detection and correction improve readout fidelity on every device and generation tested, even as the unencoded baseline improves substantially across successive hardware releases. At the same time, the value of additional redundancy depends strongly on the underlying hardware. On superconducting processors, the extra gate errors introduced by the encoding rapidly offset its benefits, whereas on trapped-ion processors the much lower gate error rates allow larger code distances to remain advantageous.
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.
Quantum error correction (QEC) enables fault-tolerant quantum computation but requires operating quantum hardware at physical error rates below code-dependent thresholds, which remains challenging for current devices. We introduce syndrome resampling, a general method that increases QEC thresholds of any decoder and suppresses logical errors without additional hardware, decoding modifications, or code-specific assumptions beyond syndrome statistics. The method exploits the fact that syndromes with low probability are likely to lead to logical failure, therefore biasing syndrome averages towards most likely syndromes effectively increases logical fidelities. We establish a direct connection between the Rényi coherent information (RCI) and powers of the syndrome probability distribution, showing that resampling syndromes according to these powers combined with maximum likelihood decoding (MLD) realizes a family of optimal thresholds associated with phase transitions in the RCI. Numerical simulations of surface codes demonstrate that syndrome resampling substantially increases thresholds for both optimal and suboptimal decoders and reduces logical error rates by up to four orders of magnitude in experimentally relevant regimes. We further show that syndrome resampling can be effectively implemented from finite data and combined with decoding-based post-selection to achieve additional gains. Finally, applying the method to existing experimental QEC data yields up to two orders of magnitude reduction in logical error rates without requiring additional measurements. Our results provide a practical and decoder-agnostic route to improved logical fidelities in near-term QEC experiments.
Quantum error correction (QEC) protects quantum computations by repeatedly measuring stabilizer syndromes and using a classical decoder to infer corrections. Modern surface-code decoders are increasingly calibration-aware: they use recent device behavior to set priors such as matching-graph edge probabilities, and these priors directly shape the selected correction. We identify the prior-update path as an overlooked integrity-critical attack surface: a poisoned prior can change decoding decisions even when the syndrome stream, logical labels, and decoder implementation are unchanged. This makes prior poisoning different from ordinary syndrome anomalies and difficult for raw syndrome-anomaly tests to detect. We introduce PriorGuard, a lightweight semantic guard that checks proposed prior changes on high-influence detector-graph entries against private syndrome-derived evidence, accepts updates only when their log-odds movement is evidence-supported, and otherwise falls back to the last evidence-supported prior. We evaluate on Stim-generated rotated surface-code memory circuits decoded with PyMatching. Prior influence is highly concentrated, with the top 10 percent of entries carrying over 60 percent of positive logical-error influence; poisoning only 6.4 percent of decoder-prior entries can raise logical error rate (LER) by 3.35x. PriorGuard recovers most attack-induced loss with a low false-positive rate and modest update-boundary overhead of about 1.7--1.8 ms and less than 0.31 MB auxiliary memory per update.
Quantum Error Correction (QEC) codes store information reliably in logical qubits by encoding them in a larger number of less reliable qubits.The surface code, known for its high resilience to physical errors, is a leading candidate for fault-tolerant quantum computing (FTQC).Logical qubits encoded with the surface code can be in different phases of their syndrome generation cycle, thereby introducing desynchronization in the system.This can occur due to the production of non-Clifford states, dropouts due to fabrication defects, and the use of other QEC codes with the surface code to reduce resource requirements.Logical operations require the syndrome generation cycles of the logical qubits involved to be synchronized.This requires the leading qubit to pause or slow down its cycle, allowing more errors to accumulate before the next cycle, thereby increasing the risk of uncorrectable errors.To synchronize the syndrome generation cycles of logical qubits, we define three policies -Passive, Active, and Hybrid.The Passive policy is the baseline, and the simplest, wherein the leading logical qubits idle until they are synchronized with the remaining logical qubits.On the other hand, the Active policy aims to slow the leading logical qubits down gradually, by inserting short idle periods before multiple code cycles.This approach reduces the logical error rate (LER) by up to 2.4× compared to the Passive policy.The Hybrid policy further reduces the LER by up to 3.4× by reducing the synchronization slack and running a few additional rounds of error correction.Furthermore, the reduction in the logical error rate with the proposed synchronization policies enables a speedup in decoding latency of up to 2.2× with a circuit-level noise model.
Quantum error correction (QEC) is the most promising route toward fault-tolerant quantum computing and, thus, useful quantum computers. QEC operates as a continuous measure-decode-correct cycle: ancilla qubits are read out, a decoder infers errors from the resulting syndromes, and corrections are applied before the next round begins. Within this loop, readout occupies a uniquely critical role, as it is the sole source of ground truth available to the decoder. Yet readout is also the slowest and most error-prone operation in the stack, with characteristics that vary across qubits and drift over time; This complexity propagates directly to the classical control hardware, and in particular to the FPGA-hosted machine-learning (ML) discriminator that must classify each analog signal into a binary syndrome outcome. Despite this central role, QEC performance has not yet been studied in depth from the perspective of readout characteristics, readout length, and their co-design with an ML discriminator.
We introduce Oraqle, an end-to-end benchmarking framework that evaluates qubit-state readout and its impact on QEC performance across real experimentally extracted qubit-state-readout datasets, state-of-the-art ML discriminators, multiple QEC codes, and hardware regimes spanning current to projected devices. Our study reveals three asymmetric findings: The measurement duration can be significantly reduced with nearly no penalty to the logical error rate; The discriminator complexity barely affects the QEC performance, as residual errors are written into device physics rather than the model; and the impact of qubit-state readout on the logical error rate is conditional on where the hardware sits in the QEC landscape, a window that widens as devices mature.