A Journey Through Modern Cosmology: From Massive Simulations to AI Scientists

Paco Villaescusa surveys the growing role of numerical simulations and AI in cosmology. Quijote and Backlight explore how much information about the Universe can be recovered beyond conventional summary statistics, including nonlinear structure and higher-order relationships. CAMELS addresses uncertainty in galaxy formation and other astrophysical processes, while DREAMS examines dark-matter predictions in the presence of baryonic effects.

The seminar then considers automated research workflows through Denario, a system that develops hypotheses, plans analyses, writes and executes code, and interprets results. It discusses possibilities opened by combining simulation libraries with coordinated AI systems, while framing these as research directions. The speaker is a computational cosmologist at the Flatiron Institute.