Real-time noise adaptive quantum error correction by multi-agent learning
Open Access
- 4 November 2026
- journal article
- Published by IOP Publishing in Quantum Science and Technology
Abstract
Quantum error correction is essential for scalable quantum computing, yet existing approaches rely on static assumptions about noise that break down in realistic hardware where error channels drift over time. We introduce a unified framework that separates QEC into two learning timescales: offline code discovery and online adaptation. Offline, Multi-Agent Reinforcement Learning (MARL) autonomously discovers complete QEC cycles as explicit quantum circuits, with separate agents responsible for encoding, syndrome extraction, and error recovery, and without prescribing a code family or circuit ansatz. Online, a lightweight adaptive layer, termed Bandit Retraining for Adaptive Variational Error Correction (BRAVE), realigns the discovered code with the drifting error channel through a transversal single-unit rotation, while a bandit controller decides from the syndrome record when an update is worth its cost. Updates are applied during operation, without decoding the logical state or retraining the MARL stack, and follow drifts on timescales of order 0.1s and longer. This yields a “discover once, adapt online” strategy that combines the flexibility of learned codes with real-time adaptation to non-stationary noise. At sufficiently high sampling rates relative to noise drift, our method reduces logical infidelity by up to 3.1-fold for qubit codes and 3.4-fold for qutrit codes compared to the strongest static alternative at each error rate, including a perfect code using two more physical units, while substantially extending robustness to noise fluctuations. These results establish a paradigm in which QEC is no longer static but dynamically optimized for realistic quantum hardware.Keywords
Funding Information
- Deutsche Forschungsgemeinschaft (ML4Q2, EXC 2004/2 – 390534769)
- HORIZON EUROPE Digital, Industry and Space (101080085, HORIZON-CL4-2021-DIGITAL-EMERGING- 02-1)
- HORIZON EUROPE Digital, Industry and Space (101113946, HORIZON- CL4-2022-QUANTUM-01-SGA)
- Bundesministerium für Forschung, Technologie und Raumfahrt (13N16149)
- Jülich Supercomputing Centre, Forschungszentrum Jülich