arXiv:2610.06059
Comparing thresholds of XZZX and rotated surface codes under spatio-temporal correlated noise
Dhruv Koul
Abstract
While independent and identically distributed noise is typically assumed in quantum computing research, complex error models such as correlated and spatio-temporal errors remain underexplored. In this paper, the performance of $\mathrm{XZZX}$ and rotated surface codes is evaluated under these complex noise models. Standard graph-based decoders struggle with the hyperedges generated by correlated errors on bias-tailored codes. To address this, we employ the Minimum Weight Parity Factor (MWPF) decoder, which natively processes syndrome hypergraphs without artificial edge decomposition. Our simulations show that under strong Z-biased circuit-level correlated noise ($η=100$), the threshold of the rotated code degrades to $\approx 0.36\%$, while the $\mathrm{XZZX}$ code drops to $\approx 0.70\%$. Furthermore, a distinct non-monotonic threshold dependence is observed for the $\mathrm{XZZX}$ code across different bias ratios. While the $\mathrm{XZZX}$ code effectively absorbs high-density Z-type hyperedges at extreme biases, it exhibits a threshold minimum at moderate biases ($η\approx 10$) because transverse X and Y errors actively bridge the diagonal strings, breaking the code's one-dimensional decoupling symmetry. These findings demonstrate that utilizing hypergraph-aware decoders is critical, as the performance of bias-tailored codes is highly sensitive to the geometric structure of spatio-temporal correlated noise.
AI Summary
The paper asks how spatially correlated faults and time-resolved circuit noise affect the thresholds of rotated and $\mathrm{XZZX}$ surface codes, particularly under Z-biased noise. It compares the codes using Stim simulations and the existing Minimum Weight Parity Factor (MWPF) decoder, which processes multi-detector hyperedges without pairwise decomposition. The noise model adds idle decoherence and adjacent/diagonal correlated faults to baseline circuit-operation errors. Its fixed overhead factors are $f_{\mathrm{idle}}=1$ and $f_{\mathrm{adj}}=f_{\mathrm{diag}}=0.25$. Thresholds are estimated by interpolating logical-error-rate crossings for distances $d=3,5,7,9$, with $d$ syndrome rounds and up to $10^7$ shots or 2,000 failures per point. At bias $\eta=p_z/p_x=100$, with $p_x=p_y$, adding these noise mechanisms reduces the estimated circuit-level threshold from $0.71\%$ to $0.36\%$ for the rotated code and from $0.89\%$ to $0.70\%$ for $\mathrm{XZZX}$. Across six biases from $\eta=1$ to $\infty$, the rotated threshold decreases from approximately $0.54\%$ to $0.34\%$. The $\mathrm{XZZX}$ threshold instead falls from $0.71\%$ to approximately $0.66\%$ near $\eta=10$, then recovers to $0.70\%$. The proposed explanation is competition between increasing hyperedge density and decreasing transverse X/Y faults that couple diagonal repetition-code chains. The stated limitations are that the correlation strengths are model assumptions rather than processor-specific measurements, and the thresholds are finite-distance numerical estimates. Larger-distance simulations and experimental validation are identified as necessary to confirm them.
Central claim may not hold: The proposed cross-chain-bridging explanation is not directly tested, and its illustrative CNOT example incorrectly propagates a control-qubit Z fault to the target. The shallow XZZX threshold minimum is reported without threshold confidence intervals or finite-size uncertainty analysis establishing that the non-monotonicity is resolved. Independent faults injected at successive ticks generate circuit-mediated temporal syndrome correlations but do not establish performance under temporally correlated physical fault occurrences. Because all reported simulations use MWPF, the paper does not demonstrate its conclusion that hypergraph-aware decoding is critical relative to graph-based alternatives. The idle-noise formula normalizes the probability of at least one fault rather than net dephasing, and using seven total ticks does not normalize the per-round overhead when faults occur only during idle ticks.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 11 pages, 8 plots, simulation codes available on GitHub