O3-Modified-Gravity-Hubble-Resolution

O3 Modified Gravity Hubble Resolution (Reviewer Seed) This repository is a review-focused seed export of the code and curated result artifacts used for: GWTC-3 / O3 dark-siren posterior-predictive scoring of a fixed modified-propagation history against an internal GR propagation baseline (selection-normalised, injection-calibrated). Follow-on Hubble-tension transfer / inference-bias analysis conditioned on the same propagation deformation. The intent is that a reviewer can: verify the headline numbers quickly (make reproduce), and rerun the main pipelines if they download the (large) public inputs. What This Repository Is Testing The model path is not “free modified gravity”; it is a fixed propagation history derived upstream and then evaluated out-of-sample on dark sirens. Conceptually: Stage A (geometry → entropy slope): reconstruct an effective horizon/entropy slope deformation μ(A) from late-time cosmology in the “minimal running-$M_\ast$” embedding implemented in src/entropy_horizon_recon/. Stage B (μ(A) → GW propagation): map to a GW propagation ratio R(z)=dLGW(z)/dLEM(z). Stage C (dark sirens): score the fixed R(z) history against an internal GR baseline using a joint posterior-predictive log score ΔLPD with explicit selection normalisation and injection-based calibration/hardening. Stage D (Hubble tension): propagate the same deformation into late-time standard-ruler inversions to quantify the induced “inference wedge” (transfer-bias) on H0. Included Contents src/entropy_horizon_recon/: core package (sirens, selection models, mapping utilities, lensing ingests). scripts/: runnable entrypoints for the O3 dark-siren hardening suite, spec-$z$ audits, and Hubble follow-on analyses. update_paper/: latest Hubble-tension manuscript source + PDF (hubble_tension_hypothesis.tex/.pdf). entropy_slope_paper/: entropy-slope letter source + PDF (context/foundation for the mapping). CQG_PAPER/: CQG-style manuscript source + PDF for the Hubble-tension implications paper (hubble_tension_cqg.tex/.pdf). CQG_DARK_SIREN/: CQG-style manuscript source + PDF for the O3 dark-siren scoring paper (dark_siren_cqg.tex/.pdf). artifacts/: curated, small “reviewer seed” outputs (tables/figures/reports). These are what make reproduce reads. Notes: Large intermediate runs live under outputs/ when generated locally; outputs/ is git-ignored by design. Large public inputs (GWTC PE files, GLADE+ indices, DESI/SDSS spectroscopy dumps, etc.) are not vendored in git. Seed Replication (Headline Verification) One command creates a timestamped reproduction folder containing report.md + summary.json with the key numbers: make reproduce This uses the curated artifacts/ bundle and does not require downloading the full public inputs. To rebuild the bundled CQG paper PDF (optional): make cqg-paper To rebuild the bundled CQG dark-siren PDF (optional): make cqg-dark-siren Output example: outputs/reviewer_seed_ /report.md outputs/reviewer_seed_ /summary.json For full reruns, see README_reproduce.md. Headline Snapshot (From artifacts/) These are the current curated “seed” headlines; make reproduce writes the same numbers to a timestamped report.md: O3 dark sirens (full score, selection-normalised): ΔLPD_tot ≈ +3.670 (ΔLPD_data ≈ +2.670), with calibrated one-sided p ≈ 0.00195 (Z ≈ 2.89) against the bundled GR-truth spectral-only null generator. Spec-$z$ override (strict shifted-sky gate; Tier A; r = 10″, K = 20000): median anchored host-weight proxy across the top-3 gate events ≈ 6.1%, with ΔLPD_tot ≈ 3.644 (non-decreasing with anchored coverage in the clean-radius regime). Hubble transfer-bias relief posterior (MC-calibrated): mean ≈ 0.246 with p16/p50/p84 ≈ 0.205 / 0.240 / 0.277. Planck lensing response refit: baseline MG projection suppresses C_L^{\\phi\\phi} near L~100 by ≈ -17.6% (median), while the constrained MG-aware response refit leaves ≈ -0.25% residual (median) and achieves χ² median ≈ 8.06 versus Planck reference ≈ 9.04. Key Result Artifacts (Vendored) Dark sirens (O3): Hardening suite (baseline reproduction, selection nuisance scan, null/permutation tests, injection checks): artifacts/o3/dark_siren_hardening_suite/report.md artifacts/o3/dark_siren_hardening_suite/tables/baseline_recompute.json artifacts/o3/dark_siren_hardening_suite/tables/calibrated_pz_table.csv Spec-$z$ “coverage maxout” audit (strict false-match gated): artifacts/o3/specz_coverage_maxout/summary.json artifacts/o3/specz_coverage_maxout/tables/false_match_gate_by_radius.csv artifacts/o3/specz_coverage_maxout/tables/override_score_best_points.csv Hubble follow-on: Transfer-bias relief posterior: artifacts/hubble/final_relief_posterior/final_relief_posterior_summary.json Planck lensing response refit (amplitude + mild ℓ-tilt): artifacts/hubble/mg_lensing_refit/tables/summary.json Joint transfer fit summary: artifacts/hubble/joint_transfer_bias_fit/tables/summary.json Artifact provenance: artifacts/manifest.json Environment Typical local setup: python3 -m venv .venv source .venv/bin/activate pip install -U pip pip install -e '.[sirens,optical_bias]' Data Source DOI Manifest (Primary) GWTC-3 PE products: DOI 10.1103/PhysRevX.13.041039 GLADE+ galaxy catalogue: DOI 10.1093/mnras/stac1443 O3 search-sensitivity injections: DOI 10.5281/zenodo.7890437 Code + reproducibility archive for the Hubble-tension manuscript: DOI 10.5281/zenodo.18635659 Notes For Reviewers This repo is intentionally structured to keep the main O3 dark-siren scoring + the Hubble transfer analysis on the same fixed propagation history. The “seed” command (make reproduce) is a headline verification tool; it does not claim to replace a full rerun with downloaded public inputs.