WB-3: Mechanism-Discrimination Test for the Wide-Binary Low-Acceleration Excess (Gaia DR4) — Frozen Preregistration, Goblin++ Analysis Chain and Laboratory Validation
This record freezes, before the Gaia DR4 release (planned 2 December 2026), the complete analysis chain for WB-3: an angle-calibrated, staged-blind test that asks whether any low-acceleration excess in wide-binary velocities (WB-1 on Gaia DR3: Gamma = 1.182 +/- 0.045, verdict INCONCLUSIVE; WB-2 protocol frozen for DR4) is a true boost of the gravitational acceleration or an artefact of the eccentricity and separation population. The eccentricity population is measured from velocity directions (the projected angle between separation and relative motion), the Newtonian expectation Gamma_N is sealed from that measurement before any velocity magnitude is examined, and the verdict statistic is B = Gamma_obs / Gamma_N. Decision rule, power gate (2,500 test pairs; 300 with both radial velocities), failure categories and known limitations are fixed in the preregistration. Contents: the frozen preregistration WB3_P0_frozen.md (SHA-256 16772a0cd8152b48e5b7f6c4936a39dffcfcb43562f0c0fc20d84e71e4cb0c57); the frozen Goblin++ and Python chain (wb3_extract.gbl, wb3_stageA.gbl, wb3_stageB.gbl, wb3_calibrate.py, wb3_stageC.gbl, wb3_verdict.py, wb1_pairs_v4.gbl) with SHA-256 hashes listed in README.md; the laboratory (wb3lab.py, wb3_synth_catalog.py, wb3_lab_rerun.sh); the six laboratory skies of 5 October 2026 (seeds 41-43, Newtonian and 17% boosted): sealed calibrations, verdicts, the 24 Goblin++ run receipts (0.1.0-alpha.23) and a spreadsheet of the results. Outcome: Newtonian skies 3/3 called correctly, boosted skies 2/3 (one inconclusive), no wrong calls; with the earlier three seeds, 12 skies and no wrong calls. A cross-check of the same six skies under Goblin++ 0.1.0-alpha.25 (built from source) reproduces every sealed and reported number; its seals and results are included. The laboratory uses synthetic skies only; no Gaia data are included. Files: WB3_zenodo_final.zip (SHA-256 afe7f86205e493e36774f627ff0fcb948d64f929c4fc5929115db7d2ac35ffe4) and WB3_P0_Preregistration_frozen.pdf, a verbatim rendering of the frozen Markdown for reading (SHA-256 7f70f5d50b7673a9a2c1fad99087b9c32ea383341273d783cff659d77f81a9c4). Engine decision recorded in the preregistration: the DR4 run uses the final Goblin++ release available before 2 December 2026, adopted only after the six laboratory skies are re-run under it and match this record; otherwise alpha.23 remains the engine of record. The frozen scripts do not change. Supplement to WB-1 (10.5281/zenodo.23108688) and its D3 addendum (10.5281/zenodo.23162815). Goblin++: https://github.com/mrgogobot/goblinpp. Claude (Anthropic) contributed as a named collaborator. The DR4 analysis will use data from the ESA mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC).
Authors
- Malin Hess
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-06
- DOI
- https://doi.org/10.5281/zenodo.23198095
- Primary Topic
- Stellar, planetary, and galactic studies
- Type
- preprint