Scalable Neural Decoders for Practical Real-Time Quantum Error Correction

2025-10-26

AI Summary

To address the prohibitively high computational complexity (O(d⁴)) of Transformer-based neural decoders—e.g., AlphaQubit—which impedes real-time quantum error correction (QEC), this work pioneers the integration of the state-space model Mamba into QEC decoding, proposing a novel lightweight and efficient neural decoder. By leveraging Mamba’s linear-time sequence modeling capability, our approach reduces decoding complexity to O(d²), enabling significantly faster inference while preserving error-correction accuracy comparable to Transformers. Evaluated on syndrome data empirically collected from Google’s Sycamore processor under realistic noise conditions, our decoder achieves a higher error threshold (0.0104 vs. 0.0097 for Transformers), demonstrating superior robustness. The method thus delivers high decoding accuracy, low latency, and strong scalability—establishing a practical, scalable decoding paradigm for large-scale fault-tolerant quantum computation.

Technology Category

• Machine Learning: Quantum Machine Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Intelligent Robots: State Estimation

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Large-scale security measurements

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 extit{AlphaQubit} have demonstrated high accuracy, the computational complexity of their core attention mechanism, which scales as $mathcal{O}(d^4)$ with code distance $d$, results in decoding speeds insufficient for practical real-time applications. In this work, we introduce and evaluate a extit{Mamba}-based decoder, a state-space model with $mathcal{O}(d^2)$ complexity. In memory experiments using Sycamore hardware data, our Mamba decoder matches the performance of its Transformer-based counterpart, providing 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, exhibiting a higher error threshold of $0.0104$ compared to $0.0097$. 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.

Problem

• Scalable neural decoders for real-time quantum error correction
• Reducing computational complexity from O(d^4) to O(d^2)
• Achieving better speed-accuracy balance in quantum decoding

Innovation

• Mamba-based decoder with O(d^2) complexity
• State-space model replacing Transformer attention mechanism
• Balances speed and accuracy for real-time quantum correction