pymatching

The quantum substrate of the Neura Parse stack — one IR across gate, neutral-atom, photonic-CV and pulse modalities; fault-tolerant promotion (surface · BB qLDPC · Gidney–Shutty cultivation); AI-augmented compilation; ed25519-signed manifests with offline-verifiable hash chains.

Geometric diagnostics for quantum error correction: A0/A1 topological classification of decoding error patterns + closed-form fault-tolerance thresholds

Production-grade Rust/Python quantum error correction decoder. 25+ decoder families: MWPM Blossom, Union-Find, BP-OSD, LDPC/qLDPC, belief-matching, CUDA/GPU batch decoding, AutoDecoder 7-tier fallback. PyMatching/Stim/Sinter compatible. Ed25519 license verification. Benchmark evidence included.

HPC assisted Surface Code threshold simulator powered by Stim, PyMatching, and Sinter for distributed quantum error correction research.

Open-source QEC decoders: Pathfinder — a direction-aware neural surface-code decoder (real-time-budget Triton kernel) whose multi-seed ensemble lowers logical error below the prior open-source GNN at operational noise rates — plus a kernel-grounded, MLE-anchored latency–LER benchmark for circuit-level qLDPC (bivariate-bicycle) codes.

A benchmark harness you can't fool: compare neural QEC decoders against minimum-weight perfect matching on identical shots, with Wilson intervals and honest three-way verdicts.

Circuit-level surface code Monte Carlo simulation with syndrome extraction, logical-failure tracking and threshold analysis.

Neural network decoders (MLP, GNN) for quantum error correction on a distance-3 surface code — benchmarked against classical MWPM, validated on real Google Sycamore hardware data.

Cloud-distributed Monte Carlo sampling for quantum error correction. Partitions Stim circuits across SkyPilot-managed spot instances, each running unmodified sinter collect, then merges results via sinter combine.

Notebook laboratory for quantum error correction, tensor-network simulation, fault-tolerant compilation, and quantum algorithms.