Bayesian Inference for Fast Scattering Glitches

Data collected by gravitational wave (GW) interferometers such as the Laser Interferometer Gravitational-wave Observatory (LIGO) is permeated by noise as a result of both environmental conditions and fundamental noise sources. Parameter estimation pipelines such as Bilby used to analyze LIGO data employ Bayesian inference, and assume that the noise in GW data is Gaussian and stationary: an assumption contradicted by the nature of non-Gaussian transient noise "glitches" prevalent within the data. Using an understanding of their physical mechanism, we have constructed a model that approximates the waveform of fast scattering glitches. We implemented this model into Bilby and tested it using a variety of fast scattering instances detected by LIGO Livingston to determine the efficacy of glitch mitigation with the model. The implementation of this model will facilitate the efficient subtraction of real fast scattering glitch instances from GW strain data, allowing for improved analysis of interesting compact binary signals in future observing runs. We demonstrate an improved mitigation of fast scattering over the GW190701 merger event from a previous LIGO-Virgo-KAGRA collaboration paper, an important step toward implementing joint inference with the model.