Physics-informed phenomenology of safety invariance in learning-based control across thermal, electrical, physiological, and financial domains — Public Companion Deposit

Statistical analysis scripts and experimental data supporting the Scientific Reports manuscript. Includes Python scripts for distributional truncation analysis (S1), boundary redirection time series (S2), intervention intensity scaling (S3), robustness analysis, and prospective real-time validation. The constraint-enforcement layer implementation is withheld under USPTO Application 19/535,932; scripts operate on the layer's observable outputs only.