Dice Question Streamline Icon: https://streamlinehq.com

Derive the null distribution of Bayes factors for precession in gravitational-wave data

Develop and characterize the distribution of Bayes factors under noise-only conditions for comparing precessing versus non-precessing waveform models, enabling calibrated significance tests of spin precession in observations such as GW231123.

Information Square Streamline Icon: https://streamlinehq.com

Background

To evaluate evidence for precession, the authors use Bayes factors and also compute the precession SNR as a complementary metric because the behavior of Bayes factors under pure noise is not currently characterized.

A principled null distribution is needed to interpret Bayes-factor values for precession across events and noise realizations.

References

Since the distribution of Bayes factors from noise alone is unknown, we additionally quantify the evidence for precession in GW231123 by computing the precession SNR, $\rho_\mathrm{p}$~\citep{Fairhurst:2019srr,Fairhurst:2019vut}.

— GW231123: a Binary Black Hole Merger with Total Mass 190-265 $M_{\odot}$ (2507.08219 - Collaboration et al., 10 Jul 2025) in Section 4, Source properties — Subsection Inference