The CAMELS Project: Cosmology and Astrophysics with Machine-learning Simulations

Проект CAMELS: космология и астрофизика с использованием симуляций машинного обучения
Greg L. Bryan, Paul Torrey, Rachel S. Somerville, Lars Hernquist, Nicholas Battaglia, Daniel Anglés‐Alcázar, Yin Li, Romeel Dave, Shirley Ho, Gabriella Contardo, Francisco Villaescusa-Navarro, Shy Genel, David N. Spergel, Annalisa Pillepich, Dylan Nelson, Desika Narayanan, Oliver H. E. Philcox, Valentina La Torre, Ana Maria Delgado, Sultan Hassan, Blakesley Burkhart, Digvijay Wadekar
2021-07-01

CAMELS projectcosmological simulationsmachine-learning algorithmsmagnetohydrodynamic simulationsstellar and AGN feedback
Abstract We present the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project. CAMELS is a suite of 4233 cosmological simulations of 25 h − 1 Mpc 3 volume each: 2184 state-of-the-art (magneto)hydrodynamic simulations run with the AREPO and GIZMO codes, employing the same baryonic subgrid physics as the IllustrisTNG and SIMBA simulations, and 2049 N-body simulations. The goal of the CAMELS project is to provide theory predictions for different observables as a function of cosmology and astrophysics, and it is the largest suite of cosmological (magneto)hydrodynamic simulations designed to train machine-learning algorithms. CAMELS contains thousands of different cosmological and astrophysical models by way of varying Ω m , σ 8, and four parameters controlling stellar and active galactic nucleus feedback, following the evolution of more than 100 billion particles and fluid elements over a combined volume of ( 400 h − 1 Mpc ) 3 . We describe the simulations in detail and characterize the large range of conditions represented in terms of the matter power spectrum, cosmic star formation rate density, galaxy stellar mass function, halo baryon fractions, and several galaxy scaling relations. We show that the IllustrisTNG and SIMBA suites produce roughly similar distributions of galaxy properties over the full parameter space but significantly different halo baryon fractions and baryonic effects on the matter power spectrum. This emphasizes the need for marginalizing over baryonic effects to extract the maximum amount of information from cosmological surveys. We illustrate the unique potential of CAMELS using several machine-learning applications, including nonlinear interpolation, parameter estimation, symbolic regression, data generation with Generative Adversarial Networks, dimensionality reduction, and anomaly detection.
1
CAMELS is designed to generate theory predictions for observables as functions of cosmology and astrophysics and to train machine-learning algorithms.
2
CAMELS provides 4,233 cosmological simulations in (25 h⁻¹ Mpc)³ volumes, including 2,184 hydrodynamic and 2,049 N-body simulations.
3
The hydrodynamic simulations use AREPO and GIZMO with baryonic subgrid physics matching the IllustrisTNG and SIMBA models.
4
The simulations evolve more than 100 billion particles and fluid elements across a combined volume of (400 h⁻¹ Mpc)³.
5
The suite varies Ωm, σ8, and four stellar- and active-galactic-nucleus-feedback parameters to represent thousands of cosmological and astrophysical models.

The CAMELS suite of cosmological (magneto)hydrodynamic and N-body simulations

Theory predictions for cosmological observables as functions of cosmological parameters and stellar and active galactic nucleus feedback, including the represented range of matter power spectra and cosmic star formation rates

Publication Details
Publication Date
2021-07-01
Journal
Publisher
ISSN
Cited by
301
Access Type
Author Information
Authors
Greg L. Bryan
Paul Torrey
Rachel S. Somerville
Lars Hernquist
Nicholas Battaglia
Daniel Anglés‐Alcázar
Yin Li
Romeel Dave
Shirley Ho
Gabriella Contardo
Francisco Villaescusa-Navarro
Shy Genel
David N. Spergel
Annalisa Pillepich
Dylan Nelson
Desika Narayanan
Oliver H. E. Philcox
Valentina La Torre
Ana Maria Delgado
Sultan Hassan
Blakesley Burkhart
Digvijay Wadekar
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%