Bipradeep Saha   (India)

bisaha @ mpia.de

Baryonic Physics and AI: Redefining Cosmological Simulations

The unprecedented observational data from instruments like the JWST has revealed gaps in our theoretical models, especially in predicting the formation and evolution of the universe's first structures. A critical hurdle in advancing these models is our limited understanding of baryonic physics—specifically, processes like star formation, feedback from active galactic nuclei (AGN), and the complex dynamics within the interstellar medium (ISM). Additionally, the exponential growth of cosmological data necessitates more efficient and scalable computational approaches to keep pace with these advancements.

In my PhD, I am addressing these challenges by developing next-generation cosmological simulations, with a core focus on improving the modeling of baryonic physics. I am working to refine the treatment of ISM and AGN feedback, both of which are pivotal for accurately predicting the formation and evolution of galaxies. These advancements will not only lead to more precise simulations but also enable the creation of robust mock catalogs that provide a meaningful comparison between theory and observation, bridging the gap between simulations and the new observational data from instruments like JWST.

A significant part of my research is dedicated to incorporating machine learning (ML) techniques to drive cosmological inferences by exploring hidden correlations between data from cosmological simulations and observation. Additionally, I will explore ML-accelerated Cosmological simulations that would help reduce computational costs.

Please feel to visit my website for more details: https://sparxastronomy.netlify.app/

Supervisor:    Annalisa Pillepich  (MPIA)