Brick by Brick: The Road to a Cohesive & Complete Story of Galaxy Formation

What is the story of galaxy formation -- when, where, and how did these vast cosmic ecosystems assemble? Armed with new cosmic photometric and spectroscopic surveys, and with James Webb data just around the corner, we are in a better position to answer this question than ever before. Yet these new data also breathe new life into a long-standing challenge in observational galaxy evolution: how do we self-consistently model all of these data? In this talk I present recent progress made using the high-dimensional galaxy SED-fitting code, Prospector. By including flexible, model-agnostic prescriptions for the complex physical processes in galaxy formation (e.g., nonparametric star formation histories), Prospector produces qualitatively new and quantitatively distinct solutions for key galaxy observables such as stellar masses and star formation rates.

I demonstrate the impact this has on our large-scale view of galaxy formation with new analyses of the stellar mass function and star-forming sequence, produced via Bayesian population modeling combined with the flexible "normalizing flow" ML technique. I argue that this Bayesian hierarchical modeling approach is the key to coherently disentangle the long-standing challenge of the strong dependence of galaxy SED-modeling on the assumed prior. I demonstrate some first steps in this direction with population models of dust attenuation and the rest-frame optical color--(mass-to-light ratio) relationship. I end by discussing how this "learn from the data" approach will be turbo-charged in the near future by lightning-fast ML-powered inference techniques combined with next-generation data (e.g., the PFS survey, JWST, and spatially resolved analysis).