Search for binary black hole sub-populations in the LIGO-Virgo gravitational wave catalog.

Details

Supervisors

de Wasseige, Gwenhaël ; Bruno, Giacomo

Faculty

Faculté des sciences

Degree label

Master [120] en sciences physiques, à finalité approfondie

Abstract

At the end of its third observation run, the LIGO/Virgo detector network has confidently detected more than 80 binary black hole (BBH) mergers, allowing the scientific community to probe for the first time the features of the BBH parameter distributions. The intrinsic properties of the binaries i.e., the masses of the two black holes and their spins can be used to study the physics of the BBH formation. Indeed, there exist various formation channels leading to binaries with different final properties. The complex features we observe in the parameter distributions suggest that the gravitational wave catalog is populated by several different populations. Yet, it is challenging to identify the origin of every BBH because the theoretical formation models predict a large range of possible parameters. The goal of this work is to search for BBH sub-populations in the gravitational wave (GW) data set using unsupervised machine learning techniques. We designed an algorithm that clusters the GW events based on the similarity of their intrinsic parameters : the primary mass, mass ratio, and effective inspiral spin. We found four clusters, and the parameter distributions of the binaries within some of them match the theoretical predictions of a formation channel. In the end, we designed the first-ever strategy, which serves as proof-of-concept, to search for an increase in coincident neutrino detection with the mergers belonging to a specific cluster. The obtained results motivate additional investigations.