Astrophysics Seminar

Cosmology is entering a transformative era. While the standard ΛCDM model explains much of what we observe, key tensions remain—including disagreement in the measurement of Hubble constant, discrepancies in the strength of matter clustering, and new 2–4σ deviations from ΛCDM reported by DESI. Resolving these puzzles—and uncovering possible new physics—may hinge on how effectively we analyze data from upcoming cosmological surveys.

Traditional analyses compress rich survey data into a few summary statistics, discarding most of the available information. I will present Field-Level Inference (FLI)—a new analysis framework that models the entire cosmological field, forward-modeling galaxy catalogs and directly comparing predictions to observations. By retaining essentially all available information, FLI enables dramatic gains in constraints on dark energy evolution and robust mitigation of astrophysical systematics that limit current analyses.

I will show the first steps towards application to real survey data, including high-resolution maps of dark matter from Dark Energy Survey survey that reveal cosmic structures with unprecedented clarity. Together, these results demonstrate how FLI can unlock the full discovery potential of next-generation cosmological surveys.