Connecting astrophysical observations and dark matter microphysics using neural simulation-based inference
Advancements in machine learning have enabled new ways of performing inference using models defined through complex simulators. In this talk, I will motivate the use of simulation-based machine learning methods for understanding the nature of dark matter using astrophysical observations, discussing advantages as well as caveats against traditional statistical techniques. I will showcase applications to several systems where the goal is to look for signatures of dark matter: ensembles of strong gravitational lenses containing dark matter substructure, astrometric observations perturbed by dark matter subhalos through weak lensing, and the kinematics of tracer stars bound to dark matter halos in satellite galaxies of the Milky Way.
|Date: ||Tuesday, 17 January 2023|
|Where: ||McGill University|
| ||Bell Room (Rutherford Physics Building, room 103)|