University of Connecticut

Events Calendar

UConn Physics Colloquium

Friday, November 12, 2021
3:30pm – 4:30pm

Storrs Campus
Gant North 20 (formerly IMS-20)

Dr. Francisco Villaescusa-Navarro, Department of Astrophysical Sciences, Princeton University

The CAMELS Project: Cosmology and Astrophysics with MachinE Learning Simulations

In this talk, I will present the Cosmology and Astrophysics with MachinE Learning Simulations --CAMELS-- project, whose main goal is to provide theoretical predictions for cosmological observables as a function of cosmology and astrophysics. CAMELS contains a collection of 4,233 cosmological simulations, 2,049 N-body simulations and 2,184 state-of-the-art (magneto-)hydrodynamic simulations employing three different codes: Gadget-III, AREPO, and GIZMO. The simulations, designed to train machine learning models, span a very wide range in the value of the cosmological and astrophysical parameters. I will illustrate the potential of CAMELS by showing a few examples of tasks that we have carried out, such as teaching neural networks to extract cosmological information and marginalize over baryonic effects at the field level, finding a new universal relation between subhalo/galaxy properties, inferring the mass of a dark matter halo from the positions and properties of its galaxies, using neural networks to reconstruct masked data, doing cosmology with one single galaxy, and teaching neural networks to learn about cosmology and astrophysics with fast radio bursts. I will conclude by discussing how CAMELS will escalate in the coming years to generate millions of hydrodynamic simulations covering gigaparsec volumes for thousands of different cosmological and astrophysical models that will be used to extract the maximum cosmological information from upcoming cosmic surveys across the entire wavelength spectrum.


Prof. Daniel Angles-Alcazar

Physics Department (primary), College of Liberal Arts and Sciences, UConn Master Calendar

Control Panel