LFCT Validation Suite: Quantitative Predictions vs Observational Data (Data Runs D thru L)

LFCT Validation Suite v3.0.5 — Data Runs D through L This archive contains nine independent Python validation runs (Data Runs D, E, F, G, H, I, J, K, L) for Light Frame Cadence Theory (LFCT), a representability-first physics framework with three modes (TD, TS, TR) sharing a cadence budget C₀ = 1/c and structural constant ε = 1/π². Each run computes quantitative predictions from LFCT's axiom set (A1–A4) plus the cluster of eight named derived principles, and compares them against published observational data from Planck 2018, AME2020, SH0ES, CODATA, McGaugh+ 2016 SPARC, BICEP/Keck, DESI 2024, and Fixsen 2009 FIRAS. Headline result: 127 quantitative tests; 91 PASS, 9 CONSISTENT, 3 MARGINAL, 0 FAIL, 3 PREDICTION, 19 STRUCTURAL. Weighted score on scored tests: 96.4%. Given the corpus-locked upstream structural constants (κ_TD = 1, κ_TS = π², κ_TR = 5/2, ε = 1/π², C₀ = 1/π, D(F) = 2F + 3, W(F) = 2π² + (2F + 3) — all derived from A1–A4 plus the cluster of eight, none fitted), no additional fitting happens downstream at any test. The three MARGINAL verdicts are the structural F2/F3 face-pair signature on the MOND acceleration scale a₀ (Run I I4f tests the pair as a unit at structural-identification tier per D26.501 + T-122 c₀ frame-split closure). Coverage: galaxy-scale dynamics (TS curvature replacing dark matter), dark-sector fractions (5/7 and 2/7 from g_s), Hubble tension as F2/F3 structural ratio (2π²+7)/(2π²+5), CMB TT production model (Score 1.88 analytical Regime A + Score 1.44 carrier-native Regime B), CMB polarisation, gravitational slip η − 1 = −1/c³, nuclear binding from Ca-40 to U-238 with the iron-peak ceiling and exact magic-number reproduction, cosmological observables (Ω_b, Ω_DE, Ω_DM, w_DE, n_s, N_eff, r = 0) at universe-as-locus, cluster-of-eight numerical instantiations. Archive contents: main README.md (audit trail + scorecard), MANIFEST.md (DOIs for 16 cited source documents: Foundation Axioms paper v3.1.0, Core trilogy v4.1.0/v4.1.0/v3.0.6, LFIS volumes 24/27/29/30/31/34/37/38, Paper Cb v1.0.0, Paper Za v1.0.1), nine LFCT_2026_DataRun{D-L}/ folders each with focused per-folder README + script + RESULTS.json, Datasets/ folder with comparison data + AME2020 fetcher script, v24_production.py for the carrier-native CMB TT regime, and GPT_AUDIT_PROMPT_v3.md for adversarial-audit transparency. All scripts use Python standard library; NumPy required only for Runs E, G, H. To verify: python3 data_run_X.py in any folder regenerates the RESULTS.json. The validation suite is built to be falsifiable: a single legitimate FAIL would propagate upstream because the structural scales are shared across regimes. v3.0.5 went through five audit cycles (GPT × 2, Gemini regular, Gemini Pro deep-think × 2) with all catches addressed honestly per the framing-bug-vs-implementation-bug discriminator; the audit chain is documented in the main README's changelog.