Galaxy Rotation Curve Analyzer: Empirical Analysis of Outer Support Requirement in SPARC Galaxies (Phases 1-104)

Overview Data analysis investigating the physical drivers of rotation curve shapes in disk galaxies using the SPARC database (Lelli, McGaugh & Schombert 2016, AJ 152, 157). This repository contains 104 phases of analysis, from initial exploration through artifact identification and reframing. Phase Summary PhasesContentStatus1-95Initial investigation of per-galaxy a₀ variation and multi-axis state lawExploratory96-100Vflat/InnerVmax ratio law for outer RC slope (r=0.85)Superseded by Phase 101101Null geometric coupling test — smooth monotonic curves reproduce r=0.833/4 null tests FAIL: ratio law is geometric artifact102Reframed: catalog-only predictors of outer mass discrepancyfgas emerges as backbone predictor (LOO R²=0.503)103Robustness battery (collinearity, bootstrap, ablation, influence, threshold)3/5 PASS, collinearity FAIL104Locked external validation — 70/30 splits, domain transfer, LITTLE THINGSExt R²=0.474, cal slope=1.01 Files Phases 101-104 (current) phase101-null-geometric-coupling.cjs/.json — Null test showing ratio law is geometric artifact phase102-residual-physics.cjs/.json — Catalog predictor search; fgas discovery phase103-robustness-battery.cjs/.json — Five robustness tests on fgas model phase104-external-replication.cjs/.json — Locked external validation (no refit) README.md — Project overview and evidence chain REPRODUCIBLE_RESULT.md — Step-by-step reproduction guide Phases 96-100 (historical, superseded) phase96-100 scripts and results — Original ratio law analysis (artifact identified in Phase 101) paper.md — Full phases 1-100 write-up (superseded) replicate_from_scratch.py — Python replication of old ratio law N45_final_dataset.csv — Quality-controlled N=45 subset Data Source All analysis uses the publicly available SPARC database (Lelli, McGaugh & Schombert 2016, AJ 152, 157). No custom or proprietary data. Note This is an ongoing empirical investigation. Results are presented as observed statistical relationships, not as established physical laws. See REPRODUCIBLE_RESULT.md for explicit caveats and scope limitations.