Illuminating the Nature of Dark Matter through Observation, Simulation and Machine Learning

Abstract

Dark matter constitutes 85% of the matter content in the universe, yet its microscopic nature remains elusive. Discovering the nature of dark matter will not only greatly further our understanding of the universe but will almost certainly shed light on what lies beyond the Standard Model of particle physics. In this thesis, I discuss the progress I have made in the hunt for dark matter by proposing direct observation strategies in the present-day universe, building simulations for dark matter energy injection imprints in the early universe, and using Machine Learning to address the unique challenges in both of these tasks. Specifically, I explore using the echoes of astrophysical radio sources to probe axion dark matter, self-consistently simulating dark matter energy injection in the era of reionization, employing simple Neural Networks to improve these early universe simulations, and utilizing Machine Learning-powered inference techniques to tackle the problem of the Galactic Center 1-ray Excess.