A multi-channel analysis testing the Kerr black hole hypothesis using 225 binary black hole merger events from GWTC-1 through GWTC-4 (O1–O4).
No statistically significant evidence for deviation from general relativity across any analysis channel. All results consistent with Kerr black holes.
| Channel | Probe | Key Statistic | Result |
|---|---|---|---|
| Multi-resolution echo search | Post-merger echoes from near-horizon reflections | σ_stack = 0.0 | No coherent echo signal |
| Model-agnostic residual analysis | Anomalous post-merger time-frequency structure | Top event: data quality issue | No beyond-GR residuals |
| Bayesian echo model comparison | Explicit ringdown+echo waveform | GW190517 ln(BF) = −6.66 | Echo model disfavored |
| Mass-scrambled null test | Population mass-dependent echo amplitude | p = 0.564 (−0.16σ) | No mass-dependent signal |
| Tidal deformability | Inspiral phase from nonzero tidal Love number | Λ̃ < 6.2 (90% CL), B₀₁ = 3091 | Decisive evidence for GR |
- 225 events analyzed: O1 (3) + O2 (8) + O3 (79) + O4 (135)
- Tidal deformability: Λ̃ = −13.7 ± 7.3, 90% upper limit Λ̃ < 6.2, Bayes factor B₀₁ = 3091
- Echo search: 8 strong candidates, 60 moderate, 152 weak, 5 none; stacked σ = 0.0
- Residual pipeline: 10 isolation forest outliers; top anomaly GW200220_124850 (data quality)
- Null test: p = 0.564, −0.16σ (222 events, 500 scrambles × 50 ε values)
- Bayesian echo: GW190517 ln(BF) = −6.66 (disfavored); GW250114 ln(BF) = 14.2 (spurious — inconsistent delay)
- Total: 973 figures, 902 JSON results, ~15 compute hours
blackhole/
├── paper/
│ └── paper.tex # LaTeX paper (~publication-quality)
├── files/
│ ├── expanded_echo_search.py # Channel 1: Multi-resolution echo search
│ ├── residual_pipeline.py # Channel 2: Model-agnostic residual analysis
│ ├── gw190517_echo_analysis.py # Channel 3a: Bayesian echo (GW190517)
│ ├── gw250114_ringdown_echo_model.py # Channel 3b: Bayesian echo (GW250114)
│ ├── novel_null_test_proper.py # Channel 4: Mass-scrambled null test
│ ├── tidal_love_pipeline.py # Channel 5: Tidal deformability
│ ├── generate_master_figure.py # Master summary figure generation
│ ├── figures/ # All generated figures (973 total)
│ │ ├── master_summary.png
│ │ ├── tidal_love/
│ │ ├── residual_pipeline/
│ │ ├── advanced/
│ │ ├── expanded/
│ │ └── events/
│ └── results/ # All JSON results (902 total)
│ ├── master_summary.json
│ ├── expanded/
│ ├── residual_pipeline/
│ ├── tidal_love/
│ └── events/
├── dashboard/ # Interactive analysis dashboard
├── README.md
├── requirements.txt
└── LICENSE
Searches for repeating post-merger signals from near-horizon reflections using ACF analysis, matched filtering with damped-sinusoid echo templates, and per-event Bayesian model selection. All 225 events are phase-aligned and coherently stacked.
Subtracts the dominant (2,2,0) QNM ringdown, then extracts 25 time-frequency features from the post-merger residual. PCA dimensionality reduction and isolation forest anomaly detection identify statistical outliers. 200 time-slide null tests calibrate each event.
Full nested sampling (dynesty) comparing noise-only, ringdown-only, and ringdown+echo models on targeted events. 20 off-source segments provide null validation.
Scans 50 Boltzmann reflectivity parameters, combining ACF measurements across the population with inverse-variance weighting. 500 random mass permutations construct the null distribution, testing for mass-dependent echo structure.
Matched-filter projection against TaylorF2 inspiral templates (3.5PN point-particle + 5PN/6PN tidal corrections). Per-event measurements combined via inverse-variance weighting. Validated with injection recovery, KS null tests, stratified analysis, and jackknife.
Strain data sourced from the Gravitational Wave Open Science Center at 4096 Hz sampling rate, 256-second windows per event, bandpass filtered 20–2000 Hz.
pip install -r requirements.txt
See requirements.txt for the full dependency list.
If you use this analysis or data products, please cite:
DHB Collaboration, "Comprehensive Test of the Kerr Black Hole Hypothesis
with 225 Gravitational-Wave Events from GWTC-1 through GWTC-4" (2026).
MIT License. See LICENSE for details.