Reproducibility Materials for Evidence-Aware Computational Screening of Hard Coatings under Incomplete and Uncertain Tribological Data: Selection and Information-Value-Guided Characterization

This record contains the revision-specific reproducibility materials for the manuscript “Evidence-Aware Computational Screening of Hard Coatings under Incomplete and Uncertain Tribological Data: Selection and Information-Value-Guided Characterization” (MRX-133500). It includes the condition-specific coating records, the independent external-consistency dataset, complete Python source, fixed seeds, generated LaTeX tables, machine-readable decision traces, all CSV outputs used in the evaluation, and the two-run SHA-256 reproducibility audit. The revised workflow separates load and environment as categorical tribosystem conditions and includes distribution, weight, dependence, observation-model, governance-threshold and attainable-EVSI sensitivity analyses, together with functional ablation. RPI was removed from the present coating architecture after an incremental-capability audit showed that it changed no coating choice, target-support result, EVPI/EVSI value or terminal action. The coating measurements were transcribed from the cited public literature; this archive does not reproduce the original experiments. Design and holdout states arise from the same declared generative model and provide internal computational verification, while the independent literature dataset supplies an external consistency test rather than experimental validation of the complete controller. This version supersedes v1 for the results reported in the revised manuscript.