TA-SR: Minimal Sufficient Reproducibility Package for Topology-Ambiguity-Aware Set-Valued Restoration

This record provides the minimal sufficient reproducibility package supporting the manuscript “Topology-ambiguity-aware set-valued restoration of DER-rich distribution networks under degraded cyber-physical observability.” The package contains 456,000 authoritative scenario-level records from four public distribution-network benchmarks, covering five random seeds, twelve evaluated methods, four candidate budgets, and eight switch-telemetry degradation mechanisms. It also includes the source code, frozen experiment configurations, random seeds, analysis and plotting scripts, balanced AC power-flow audits, DER-availability and telemetry-degradation sensitivity results, topology-particle sensitivity results, cross-fitted AC-risk calibration data, and same-hardware inference-timing records. The package supports two levels of reproducibility: (1) exact regeneration of the principal aggregate analyses, statistical summaries, tables, and data figures from the retained scenario-level records; and (2) computational reruns of scenario generation, model training, candidate generation, ambiguity-aware ranking, and physical validation using the supplied code and configurations. To respect third-party licensing conditions, the original pandapower and SimBench network files and serialized topology tables are not redistributed. The package instead provides the network identifiers, pinned software versions, experiment configurations, and reconstruction code. The study networks can be reconstructed using pandapower 3.4.0 and SimBench 1.6.2 from their official distributions. Package integrity is documented through a SHA-256 manifest. The archived package contains 20 authoritative scenario-result files and has passed 19 automated tests. A clean reproduction run regenerated the principal analysis outputs with exact SHA-256 agreement.