Inferring 2D exsitu distribution of galaxies with AI

The proposed research project aims to gain crucial insights into the stellar assembly of galaxies using advanced machine learning techniques. Most galaxies in the Universe are not isolated systems; they grow primarily through interactions and mergers with other galaxies. While we know that mergers significantly impact galaxy evolution, quantifying the contribution of accretion to a galaxy’s total stellar mass remains a challenge. By training state-of-the-art machine learning models, such as diffusion models, on cosmological simulations, we aim to infer properties of galaxies that are not directly observable from their observational features. Specifically, we seek to predict for the first time the distribution of ex-situ versus in-situ stellar mass in nearby galaxies, providing a detailed 2D projection of these components. This project, which is part of the intern’s PhD thesis, will leverage machine learning to uncover hidden aspects of galaxy evolution. The project will offer a unique perspective on the merging history of galaxies and demonstrate the potential of simulation-based inference in astrophysical research. The insights gained are expected to significantly enhance our understanding of galaxy formation and evolution, with the potential to lead to a scientific publication.

Faculty Supervisor:

Laurence Perreault-Levasseur;Yashar Hezaveh

Student:

Partner:

Institute of Astrophysics of the Canary Islands

Discipline:

Physics

Sector:

Education

University:

Université de Montréal

Program: