Scalable neural decoders for practical real-time quantum error correction
Abstract
Real-time, scalable, and accurate decoding is a critical component for realizing a fault-tolerant quantum computer. While Transformer-based neural decoders such as AlphaQubit have demonstrated high accuracy, the computational complexity of their core attention mechanism, which scales as 𝒪(d4) O ( d 4 ) with code distance d , results in decoding speeds insufficient for practical real-time applications. In this work, we introduce and evaluate a Mamba -based decoder, a state-space model with 𝒪(d2) O ( d 2 ) complexity. In memory experiments using Sycamore hardware data, our Mamba decoder matches the performance of its Transformer-based counterpart, demonstrating that its superior efficiency does not come at the cost of performance. Crucially, in simulated real-time scenarios that account for decoder-induced noise, the Mamba decoder significantly outperforms the Transformer, with the performance gap increasing across the tested code distances up to d = 9. These results demonstrate that Mamba decoders offer a compelling balance between speed and accuracy, making them a promising architecture for scalable, real-time quantum error correction.
Related
- A Cross-Platform Analysis of Practical Quantum Error Correction Codes
- Graph Neural Post-selection for Quantum Error Correction
- Scalable Message-Passing Quantum Graph Neural Networks in the Weisfeiler-Leman Hierarchy
- Local decoders for fault-tolerant quantum computation and translation-invariant stabilizer codes
- Certified Non-Degenerate Maximum-Likelihood Benchmarking of Quantum LDPC Decoders
BibTeX
@article{lee2026scalable,
title = {{Scalable neural decoders for practical real-time quantum error correction}},
author = {Changwon Lee and Tak Hur and Daniel K. Park},
journal = {npj Quantum Information},
year = {2026},
month = oct,
doi = {10.1038/s41534-026-01403-y},
url = {https://www.nature.com/articles/s41534-026-01403-y},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Scalable neural decoders for practical real-time quantum error correction
AU - Changwon Lee
AU - Tak Hur
AU - Daniel K. Park
PY - 2026
DA - 2026/10/28
JO - npj Quantum Information
DO - 10.1038/s41534-026-01403-y
UR - https://www.nature.com/articles/s41534-026-01403-y
AB - Real-time, scalable, and accurate decoding is a critical component for realizing a fault-tolerant quantum computer. While Transformer-based neural decoders such as AlphaQubit have demonstrated high accuracy, the computational complexity of their core attention mechanism, which scales as 𝒪(d^4) O ( d 4 ) with code distance d , results in decoding speeds insufficient for practical real-time applications. In this work, we introduce and evaluate a Mamba -based decoder, a state-space model with 𝒪(d^2) O ( d 2 ) complexity. In memory experiments using Sycamore hardware data, our Mamba decoder matches the performance of its Transformer-based counterpart, demonstrating that its superior efficiency does not come at the cost of performance. Crucially, in simulated real-time scenarios that account for decoder-induced noise, the Mamba decoder significantly outperforms the Transformer, with the performance gap increasing across the tested code distances up to d = 9. These results demonstrate that Mamba decoders offer a compelling balance between speed and accuracy, making them a promising architecture for scalable, real-time quantum error correction.
SN - 2056-6387
ER -