| Abstract: |
In this seminar I will present the CAMELS project, whose aim is to provide theoretical predictions for cosmological observables as a function of cosmology and astrophysics. Containing a set of 4,233 simulations, both N-body and state-of-the-art hydrodynamic simulations, it is designed as a large dataset to train machine learning models. I will present some of the results obtained so far, including the finding of a universal relation in subhalo properties, how neural networks can extract cosmological information while marginalizing over baryonic effects at the field level, and the usage of graph neural networks to weight the Milky Way and Andromeda. In view of the CAMELS full data release scheduled on January 6th 2022, I will describe some specifications of CAMELS data and the science that can be done with it. |
