Machine Learning Supermassive Black Holes

Supermassive black holes are beautiful confirmation of our laws of physics as well as a crucial ingredient in our understanding of how galaxies form and evolve. Until now, black holes in galaxies have been measured by assuming that galaxies have simple geometries and are in a steady state. However, real galaxies have more complex shapes and vary over time.

Ideally one would like to be able to fit realistic N-body simulation of galaxies directly to the data. However, so far this has proven computationally unfeasible. Machine learning has become pervasive in our daily lives, thanks to recent algorithmic breakthrough and has the potential of solving the problem.

In this project, the student will explore how machine learning and AI can be used for a new approach to the dynamical modelling of galaxies and to measure masses of supermassive black holes. The student will then extract the black hole masses from the stellar kinematics of a sample of galaxies. The measured masses will be used to update black holes scaling relations and try to improve our understanding of the role of black holes in galaxy evolution.

Altough this project is studying supermassive black holes, it can have an impact well beyond this scientific topic. In fact, we aim at studying the feasibility of using machine learning as a substitute for dynamical modelling methods that have been based on similar principles for over a century. Even more generally, the project will investigate the effectiveness of using machine learning to fit general models to data. If successful, this idea can have broad applications to other scientific domains.