Speaker
Description
Modern cosmological surveys are specifically designed to characterise the dark energy equation of state and dark matter power spectrum, and place tight constraints on the neutrino mass. Unfortunately, a large part of the data is currently discarded because of the lack of robust theoretical modelling on small scales. The main uncertainties are associated with baryonic physics and galaxy formation. In my talk, I will introduce a novel hybrid framework, capable of optimally analysing the incoming sky surveys, leveraging a hybrid approach that mixes post-processed N-body simulations and machine learning. The post-processed baryonification technique allows the accurate and fast prediction of multiple cosmic fields in a wide range of astrophysical scenarios, by flexibly painting galaxies and gas on top of the dark matter. Artificial neural networks are used to train fast emulators of 2pt and higher-order statistics as a function of cosmological and astrophysical scenarios. The emulators can be seamlessly embedded in a Bayesian pipeline for multi-probe analyses, which can include cosmic shear, galaxy clustering, diffuse X-ray, thermal Sunyaev-Zel’dovich effect and their cross-correlation.