Robust cosmological inference from galaxy clustering and weak lensing using cosmological simulations

Cross-correlations between imaging and redshift surveys of galaxies and high-resolution observations of the CMB promise to shed light on the physical nature of dark matter and dark energy in the coming decade. One of the main factors limiting the precision and accuracy of cosmological constraints coming from these measurements will be our understanding of the physics of galaxy formation. In this talk, I will present a roadmap for leveraging cosmological simulations to provide highly flexible models for this physics, paving the way for robust cosmological inference. First, I will show how data-driven models of galaxy formation and evolution combined with contemporary machine learning techniques can be used as robustness tests for complex cross-correlation analyses, with case studies from the Dark Energy Survey and the Dark Energy Spectroscopic Instrument. I will then discuss recent progress on combining perturbative models of structure formation with N-body simulations in order to obtain robust predictions for galaxy clustering and weak lensing, and describe how similar models might be used to confront a variety of future observations.