Neural vs. classical decoders for the surface code: benchmarking MWPM, CNN, and Transformers on simulated and real IBM quantum hardware.
-
Updated
Jul 3, 2026 - Jupyter Notebook
Neural vs. classical decoders for the surface code: benchmarking MWPM, CNN, and Transformers on simulated and real IBM quantum hardware.
A benchmarking framework for surface-code decoders: MWPM, union-find and belief propagation compared on accuracy, runtime and memory.
Reproduction of 'Testing the Accuracy of Surface Code Decoders' (arXiv:2311.12503): exact MWPM vs optimal maximum-likelihood decoder.
Source-available Rust/Python quantum error correction decoder package with MWPM, Union-Find, BP-OSD, LDPC/qLDPC, CUDA batch decoding, PyMatching/Stim validation, and reproducible benchmark artifacts.
Add a description, image, and links to the mwpm topic page so that developers can more easily learn about it.
To associate your repository with the mwpm topic, visit your repo's landing page and select "manage topics."