Topological noise folding for scalable quantum error mitigation

This record contains the official research code, datasets, and reproducibility artifacts for the project "Topological Noise Folding for Scalable Quantum Error Mitigation". Topological Noise Folding (TNF) is a neuro-symbolic quantum circuit optimization framework that combines ZX-calculus representations with graph neural networks and reinforcement learning to discover noise-suppressing diagrammatic rewrites. The framework enforces Pauli-flow constraints during optimization, ensuring deterministic circuit extraction with empirical polynomial-time scaling. The repository includes: • Source code for the Graph Transformer optimization framework • Benchmark results across 81 quantum circuits • Hardware validation results from the IBM Fez 156-qubit superconducting quantum processor • Reproducibility scripts used to generate all figures and results reported in the manuscript Key results: • TNF outperforms heuristic compilers on 98.8% of circuits • Average T-count reduction: 19.7% • Maximum reduction observed: 40.3% • Hardware validation shows +2.39% fidelity improvement on a QAOA circuit This archive accompanies the research manuscript: "Topological Noise Folding for Scalable Quantum Error Mitigation".