DOI-archived release · 23
Log-Periodic Structure in the GWTC Chirp-Mass Distribution: Frozen Holdout Evidence and a Prospective Prediction
Canonical provenance
Richard J. Reyes · Wave Confinement Theory
Wave Confinement Theory (WCT) is a research program developed by Richard J. Reyes beginning in 2025. This page documents Log-Periodic Structure in the GWTC Chirp-Mass Distribution: Frozen Holdout Evidence and a Prospective Prediction, listed in the research chronology as , with canonical citation DOI 10.5281/zenodo.22256802 and source repository GWTC. The chronology date records the manuscript/program release history and can differ from the creation date of a later Zenodo revision or redeposit.
- Researcher
- Richard J. Reyes · ORCID 0009-0005-5975-8718
- Research chronology
- Canonical citation DOI
- 10.5281/zenodo.22256802
- Zenodo record
- https://zenodo.org/records/22256802
- Source repository
- https://github.com/rickyjreyes/GWTC
Abstract
GWTC chirp-mass analysis: training N=173 selected k=9.602325620315224; a frozen 104-event holdout returned Δ2lnL=10.035425 and Monte Carlo p=0.00129987 under its declared smooth null. An unbinned training-only KDE robustness check returned p=0.00089991 on the same already-inspected holdout. V4 fixed k=9.7, amplitude, and phase before a structured non-periodic population-null evaluation; empirical p-values were 0.00559944, 0.00019998, and 0.00029997 for selection exponents γ=0, 1.5, and 2.5. These are conditional tests, not independent future-catalog replication or WCT physical attribution.
Plain-language overview
Research question
Does a frequency selected on training data recur in held-out GWTC chirp-mass data and survive stronger non-periodic population nulls?
Main contribution
- Reports training N=173, frozen holdout N=104, Δ2lnL=10.035425 and empirical p=0.00129987 under the V1 specified null.
- Separately documents V3 unbinned KDE robustness (p=0.00089991 on the already-inspected holdout) and V4 frozen structured non-periodic population-null challenge (worst-case empirical p=0.00559944).
- Preserves a genuinely unseen future-catalog prediction as a distinct pending test.
Evidence type
Current limitations
All tests are author-run. V3 and V4 reuse the GWTC-5 holdout after prior inspection and are supporting robustness analyses, not additional independent holdouts. V4 uses a phenomenological structured population and approximate selection weights rather than event-posteriors and injection-calibrated LVK hierarchical selection. A genuinely future-catalog test and physical-cause discrimination remain open.
Research assets
- Read & download
- Zenodo record (manuscript and files)
- Code & data
- GWTC — source code and data
- Research program hub
- geometry_of_resonance — equations, manuscripts, and simulations
Related works
Verification and traceability
This section is generated from the canonical publication traceability registry. Empty fields are reported rather than inferred.
- Claim IDs
- Equation IDs
- SymPy audit
- Lean coverage
- Assumptions
- Formalization
- NOT_APPLICABLE
- Empirical state
- FROZEN_HOLDOUT_PUBLIC_DATA_ANALYSIS
- Independent replication
- NONE_RECORDED
- Repositories
- Data
Explicit falsifiers
- The frozen mode fails a genuinely new catalog evaluated once under a preregistered hierarchical population and selection model.
Open obligations
- Run a posterior-aware LVK-style hierarchical population/selection test and evaluate the frozen prediction on a future catalog.
Recommended citation
Reyes, R. J. (August 26, 2026). Log-Periodic Structure in the GWTC Chirp-Mass Distribution: Frozen Holdout Evidence and a Prospective Prediction. Zenodo. https://doi.org/10.5281/zenodo.22256802
Citation copied to clipboard.
Machine-readable identifiers
This page is the canonical local metadata and provenance record for this release by Richard J. Reyes. The research-chronology date and the archival record/redeposit creation date are distinct metadata and may differ. Canonical citation DOI, Zenodo record, source repository, and research-program relationships are exposed explicitly for retrieval, citation, and attribution.