GCV: Numerical selection of Lambda_eff (Λ_eff) on the GR-exact branch

GCV: Numerical selection of Lambda_eff (Λ_eff) on the GR-exact branch

v6.0.0 release of the “GCV Numerical Λ_eff selection” reproducibility bundle.

Fixed inputs (held fixed in this release):
• Fixed late-time selection posterior P_sel(x) for x ≡ Λ/Λ_obs (bundled selection posterior release; treated as input).
• UV anchoring held fixed (bundled UV mapping workbook; discretuum gap proxy x_min,UV).
• Single positive-Λ measure proxy on the accelerating branch (Press–Schechter–motivated erfc form), anchored at x = 1.

New in this release:
• Explicit Λ_obs anchoring and absolute-Λ outputs: in addition to x ≡ Λ/Λ_obs posteriors, this release outputs posteriors for Λ in m⁻² and cm⁻² for a declared Λ_obs anchor (anchor database bundled).
• Capability model upgrade and comparison modes:
  – “sn_only”: minimal SN-like Fisher capability model (legacy mode; direct comparison to v5-style capability).
  – “sn_ap”: SN capability augmented with a ruler-free AP geometry capability track (to reduce Ω_m–Λ degeneracy reliance).
• Tightening-attribution exports: stage-by-stage posteriors and weight components to show which pipeline components drive reweighting (selection input → UV anchoring → measure proxy → capability conditioning).
• Sensitivity dashboard: structured stress tests of key capability assumptions (survey reach, SN scatter/sample size, AP precision scaling, σ_{Ωm} envelope), with summary tables and figures.
• Public, versioned scenario interface: named capability scenarios (baseline/optimistic/conservative + envelope scenarios) defined in configs/, with exported “scenario cards” for auditability and citation.

Reproducibility:
• Single reproduce script: lambda_posterior_step6_v6p0p0_reproduce.py
• Convenience runners: RUN_TEST_WINDOWS.bat and RUN_TEST_UNIX.sh
• Versioned outputs written to: results/ (CSV+JSON), figures/, and PUB/ exports
  – outputs are separated by capability mode under results/sn_ap, results/sn_only, figures/sn_ap, figures/sn_only, etc.
• MANIFEST_SHA256 provides file hashes for archival integrity.
• requirements.txt and requirements_pinned.txt provided; each run writes an environment report JSON for debugging/version tracing.

Interpretation note:
Capability is defined on the accelerating branch (x > 0). Therefore, capability-conditioned posteriors are conditional on x > 0. The unconditional sign probability remains that of the fixed late-time selection posterior P_sel(x).

Release date: 2026-02-25

Project homepage: https://johansson.digital
Project overview hub: https://github.com/gcv-framework/gcv-vacuum-energy