arXiv:2609.26958
Soft decoding for quantum LDPC codes with experimental validation
Arda Aydin, Edwin Tham, Nicolas Delfosse, Min Ye
Abstract
The decoder is a critical component of a fault-tolerant quantum computer, computing corrections based on parity-check measurements performed throughout the computation. A soft decoder supplements its output with a confidence score which, when used alongside post-selection, can substantially improve logical performance. We introduce a soft beam search decoder for quantum low-density parity-check codes that uses internal decoder data as a confidence metric, removing the need for extra computation. We perform circuit-level simulations of five quantum LDPC codes relevant for superconducting and trapped ion architectures equipped with our global soft decoder and we obtain up to $580\times$ logical-error suppression at a physical error rate of $10^{-3}$ while rejecting only $0.1\%$ of shots. Then, we simulate an error detected measurement, which is a core logical operation of the walking cat architecture, using a streaming version of soft beam decoder and we achieve up to $210\times$ error suppression while increasing the rejection probability by only $0.5$ percentage points. Finally, we revisit recent quantum LDPC code memory experiments on trapped ions, demonstrating that our soft decoder doubles the logical qubit lifetimes at the price of a mean rejection rate of $2.6\%$--$5.6\%$ per syndrome round, bringing all five codes into the beyond-breakeven regime.
AI Summary
Can qLDPC decoders provide useful confidence estimates without additional decoding passes? This paper uses the cumulative number of belief-propagation iterations across beam-search paths as a reliability metric: shots exceeding a cutoff are rejected. The cutoff also bounds worst-case BP work. A hybrid rule applies the existing two-round logical-error criterion (2R-LEC) only to shots with intermediate iteration counts; a sliding-window extension rejects shots if any window exceeds the cutoff. Circuit-level simulations cover five qLDPC codes. At physical error rate $p=10^{-3}$, iteration-cutoff post-selection suppresses logical errors by $124\times$ for $[[90,8,10]]$ and $78\times$ for $[[144,12,12]]$, rejecting $0.1\%$ of shots. The hybrid reaches $580\times$ suppression for $[[54,2,10]]$ at the same rejection rate. At the reported operating points, it outperforms 2R-LEC while using $32\%$–$65\%$ of its mean BP iterations. Sliding-window simulations of error-detected logical measurements obtain up to $210\times$ suppression for a $0.5$-percentage-point increase in restart probability. Re-decoding existing trapped-ion memory data for five codes more than doubles fitted conditional logical lifetimes at mean rejection rates of $2.6\%$–$5.6\%$ per syndrome round, without new experiments. All five central estimates exceed the physical-lifetime reference, but one code’s uncertainty interval still overlaps it. The confidence metric is empirical, and cutoff-only suppression is less effective for the tested $[[102,22,9]]$ code. Results concern post-selected ensembles, with thresholds calibrated to rejection budgets; proposed benefits to magic-state factories and larger computations are not directly demonstrated.
Error found: The abstract associates its no-extra-computation confidence method with 580× suppression without explaining that this result uses the hybrid rule, which requires an additional decoder pass on selected shots; iteration cutoff alone achieves 346× at the same rejection rate.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.