Modelling noise in gravitational-wave observatories with transdimensional models

TL;DR

This paper presents a novel transdimensional Bayesian method for modeling noise in gravitational-wave detectors, improving noise characterization and affecting astrophysical parameter estimation.

Contribution

It introduces a flexible Bayesian noise model using power laws, Lorentzians, and shapelets, implemented with Bilby, enhancing noise fits for LIGO and Virgo data.

Findings

Improved fit of noise spectral densities over existing models

Observed up to 7% shifts in credible interval boundaries for some parameters

Framework ready for systematic deployment in gravitational-wave inference

Abstract

Modelling noise in gravitational-wave observatories is crucial for accurately inferring the properties of gravitational-wave sources. We introduce a transdimensional Bayesian approach to characterise the noise in ground-based gravitational-wave observatories using the Bayesian inference software $\texttt{Bilby}$. The algorithm models broadband noise with a combination of power laws; narrowband features with Lorentzians; and shapelets to capture any additional features in the data. We show that our noise model provides a significantly improved fit of the LIGO and Virgo noise amplitude spectral densities compared to currently available noise fits obtained with on-source data segments. We perform astrophysical inference on well-known events in the third Gravitational-Wave Transient Catalog using our noise model and observe shifts of up to $7\%$ in the $90\%$ boundaries of credible intervals for some parameters. We discuss plans to deploy this framework systematically for gravitational-wave inference along with possible areas of improvement.