Code and data for: Five-qubit graph-state subcodes: enumeration, decoding and structural prediction

This repository contains the complete datasets, executable code, and model predictions accompanying the manuscript "Five-qubit graph-state subcodes: enumeration, decoding and structural prediction" (submitted to Quantum Information Processing). Contents of the archive (ESM_2.zip):- data/core/canonical_dataset.csv: All 22,568 distinct stabilizer constructions, structural descriptors, exact failure counts, and rational targets.- data/core/lc_signed_certificates.json: Signed local-Clifford equivalence certificates for all 132 distance-three constructions.- data/ml_features.csv: Prepared feature and target table with stable construction IDs.- results/core/: Complete enumeration results, temporal identities, and independent Stim validation checks.- results/ml/: Model configurations, deterministic fold assignments, per-row predictions, and full regression results.- code/: Self-contained Python scripts for data generation, regression, metric verification, and figure generation. The analysis employs exact integer-numerator arithmetic for all target probabilities, ensuring strict reproducibility without floating-point accumulation errors. The code is provided under the MIT License, and the data is provided under the CC-BY-4.0 License. Online Resource 1 (ESM_1.pdf) contains the supplementary mathematical derivations, full regression tables, and reproduction protocol.