Looking back from the stellar graveyard: inferring the origins of binary black holes with gravitational waves

Colloms, Storm (2026) Looking back from the stellar graveyard: inferring the origins of binary black holes with gravitational waves. PhD thesis, University of Glasgow.

Abstract

Gravitational waves (GWs) have transformed our understanding of binary black holes (BBHs), with hundreds of observations now published since the first direct detection in 2015. Measuring the population properties of many BBHs allows us to statistically infer the of the lives and deaths of their progenitors, which are subject to many overlapping uncertainties. The first source of these uncertainties arises from observational limitations: BBHs likely originate from massive, potentially low-metallicity stars, which are short-lived and hard to observe. Second, predicting the evolution of massive stars is limited by our theoretical understanding of multi-scale stellar evolutionary processes, which for BBH progenitors is further complicated by binary and dynamical interactions. Third, understanding the astrophysics over a population of BBH progenitors is limited by the computational expense of simulating the detailed physics of their progenitors using population synthesis models. The growing number of GW observations offer unique insight into these uncertainties, allowing us to understand the evolutionary pathways that lead to merging BBHs.

In this thesis, I investigate the origins of BBHs with new complementary modelling frameworks, exploring correlations between observed masses and mass ratios, and developing emulators of population synthesis simulations.

Motivated by evidence that BBHs prefer to merge with similar masses, I develop parametric population models to infer whether the minimum mass of the smaller black hole varies with the mass of the larger black hole. Using GW data from hundreds of observations, I show that big black holes can merge with the smallest black holes, contrary to expectations from previous studies of the BBH mass ratio distribution. I find that the interpretation of population outliers, such as unequal mass ratio observation GW190814, depends on the construction of our population models. This demonstrates how analysing GW observations using new population models furthers our understanding BBH observations and astrophysics

Comparing the observed BBH population to population synthesis simulations allows for constraints on astrophysical parameters governing their evolution, and the relative contribution from different formation channels. These simulations are computationally expensive and rely on a wide range of assumptions, therefore achieving robust predictions from population synthesis remains a key challenge in GW astronomy. I demonstrate the effectiveness of normalising flows, a type machine learning algorithm, to interpolate predictions from population synthesis and overcome the necessity of a dense grid of simulations. This framework enables previously intractable constraints over a continuous range of parameters governing the physics of BBH evolution, and probe degeneracies between key assumptions in population synthesis simulations.

Extending this framework to ensure consistency between the emulated populations and the observational selection effects, I find that differences in how selection effects are calculated lead to large discrepancies in inference results with O(100) observations. These discrepancies arise from small differences between the emulated populations and the underlying population synthesis models, indicating that more accurate emulation of population synthesis is necessary avoid systematic bias in resulting astrophysical constraints. This is particularly important as we analyse an increasing number of observations, which will drive increasing differences between constraints under these two methods.

Showing how these analyses are subject to shrinking statistical and persistent systematic uncertainties with growing GW catalogues, I highlight the future work necessary for robust BBH population inference. Overcoming these systematics will enable the application of new population models, including more flexible parametric models of correlated distributions, and population synthesis models exploring the correlations between more astrophysical parameters and formation channels. The development of robust inference frameworks will allow us to confidently constrain multiple uncertain parameters in BBH evolution, and develop trustworthy simulations of BBH progenitor populations. As we advance further into the era of GW astronomy, these analysis provide crucial steps towards piecing together where and how BBHs form.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QB Astronomy; Q Science > QC Physics
• Colleges/Schools: College of Science and Engineering > School of Physics and Astronomy
• Supervisor's Name: Berry, Dr. Christopher and Veitch, Professor John
• Date of Award: 2026
• Depositing User: Theses Team
• Unique ID: glathesis:2026-86285
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 30 Oct 2026 11:15
• Last Modified: 30 Oct 2026 11:37
• Thesis DOI: 10.5525/gla.thesis.86285
• URI: https://theses.gla.ac.uk/id/eprint/86285

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Pre-print