Exploring the astrophysical origins of binary black holes using normalising flows

TL;DR

This paper uses normalising flows to efficiently emulate multiple binary black hole population synthesis models, enabling detailed analysis of formation channels and evolution using gravitational-wave data.

Contribution

It introduces a novel application of normalising flows to emulate complex population-synthesis models, reducing computational costs and facilitating interpolation between different formation scenarios.

Findings

Normalising flows accurately emulate population-synthesis models.

The method enables measurement of formation channel branching ratios.

It allows detailed analysis of binary evolution from gravitational-wave data.

Abstract

The growing number of gravitational-wave detections from binary black holes enables increasingly precise measurements of their population properties. The observed population is most likely drawn from multiple formation channels. Population-synthesis simulations allow detailed modelling of each of these channels, and comparing population-synthesis models with the observations allows us to constrain the uncertain physics of binary black hole formation and evolution. However, the most detailed population-synthesis codes are computationally expensive. We demonstrate the use of normalising flows to emulate five different population synthesis models, reducing the computational expense, and allowing interpolation between the populations predicted for different simulation inputs. With the trained normalising flows, we measure the branching ratios of different formation channels and details of binary stellar evolution, using the current catalogue of gravitational-wave observations.