Low-Latency Belief-Propagation Decoding of Quantum LDPC Codes with Reinforcement-Learned Scheduling
Abstract
Belief propagation (BP) is attractive for quantum low-density parity-check (QLDPC) decoding because of its low complexity and low latency, but short cycles and degeneracy can limit its error-correction performance. Reinforcement-learned sequential scheduling (RL-S) substantially improves this performance by adapting the variable node (VN) update order, although Q-table storage and heap-based selection introduce additional overhead. We propose bucket RL-S (B-RL-S), which quantizes learned Q-values into integer priorities and schedules VN updates using dynamically maintained first-in, first-out buckets. Without retraining or changing the BP message updates, B-RL-S reduces policy storage and refreshes only priorities affected by local residual syndrome changes. A fixed number of priority levels enables constant-time VN selection and queue relocation. For the $[[288,12,18]]$ bivariate bicycle code over the depolarizing channel at p=0.05, B-RL-S with eight priority levels achieves approximately one fifth the block error rate (BLER) of flooding BP followed by order-10 ordered-statistics decoding. The respective caps are $1000$ sequential iterations for B-RL-S and $1000$ flooding iterations for the BP stage of the reference decoder. These results highlight strong error correction with inexpensive scheduling and no additional post-processing stage.
Related
- Conflict-Free Color-Clustered Sequential Belief-Propagation Decoding of Quantum LDPC Codes via Reinforcement Learning
- Low-Latency LDPC Decoding Based on Consensus and Return Residual Belief Propagation
- Hardware-Efficient Reinforcement-Learned Scheduling for Quantum LDPC Decoders
- Learned Parallel Bit-Flipping Sequential Belief Propagation Decoding of Quantum LDPC Codes
- Belief Propagation-Informed Tesseract Decoding of Quantum LDPC Codes
BibTeX
@misc{moradi2026lowlatency,
title = {{Low-Latency Belief-Propagation Decoding of Quantum LDPC Codes with Reinforcement-Learned Scheduling}},
author = {Mohsen Moradi and Ghilas Aissou and Taejoon Kim and Remi A. Chou and David G. M. Mitchell},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.15857},
eprint = {2610.15857},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.15857},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Low-Latency Belief-Propagation Decoding of Quantum LDPC Codes with Reinforcement-Learned Scheduling
AU - Mohsen Moradi
AU - Ghilas Aissou
AU - Taejoon Kim
AU - Remi A. Chou
AU - David G. M. Mitchell
PY - 2026
DA - 2026/10/11
PB - arXiv
DO - 10.48550/arxiv.2610.15857
UR - https://arxiv.org/abs/2610.15857
AB - Belief propagation (BP) is attractive for quantum low-density parity-check (QLDPC) decoding because of its low complexity and low latency, but short cycles and degeneracy can limit its error-correction performance. Reinforcement-learned sequential scheduling (RL-S) substantially improves this performance by adapting the variable node (VN) update order, although Q-table storage and heap-based selection introduce additional overhead. We propose bucket RL-S (B-RL-S), which quantizes learned Q-values into integer priorities and schedules VN updates using dynamically maintained first-in, first-out buckets. Without retraining or changing the BP message updates, B-RL-S reduces policy storage and refreshes only priorities affected by local residual syndrome changes. A fixed number of priority levels enables constant-time VN selection and queue relocation. For the $[[288,12,18]]$ bivariate bicycle code over the depolarizing channel at p=0.05, B-RL-S with eight priority levels achieves approximately one fifth the block error rate (BLER) of flooding BP followed by order-10 ordered-statistics decoding. The respective caps are $1000$ sequential iterations for B-RL-S and $1000$ flooding iterations for the BP stage of the reference decoder. These results highlight strong error correction with inexpensive scheduling and no additional post-processing stage.
ER -