Physics ∩ ML

a virtual hub at the interface of theoretical physics and deep learning.

18 May 2022

Simulation-based inference for astrophysical dark matter searches

Siddharth Mishra-Sharma, MIT


In this talk, I will motivate the use of simulation-based machine learning methods for new physics searches, in particular for understanding the nature of dark matter, using astrophysical observations. After showcasing several applications and discussing advantages as well as caveats against traditional techniques, I will spend the bulk of the talk describing a study of gamma-ray data from the center of the Milky Way where the goal is to characterize a potential signal of dark matter – the so-called Galactic Center Excess. While emphasizing the modeling and inference challenges associated with this task, I will discuss how leveraging machine learning methods offers a path towards resolving the long-standing puzzle of the origin of the signal.