Horizon-AGN

Horizon-AGN
Name	Horizon-AGN
Type	cosmological hydrodynamical simulation
Institution	CEA Saclay
Start	2009
Volume	100 Mpc/h
Resolution	1024^3 dark matter particles
Code	RAMSES
Primary publication	Dubois et al. (2014)

Horizon-AGN is a large cosmological hydrodynamical simulation project designed to model the formation and evolution of galaxies, black holes, and large-scale structure within a ΛCDM framework. It couples N-body dynamics with adaptive mesh refinement hydrodynamics to follow baryons and dark matter across cosmic time, and incorporates subgrid prescriptions for star formation, feedback, and active galactic nuclei. The project has been used by research groups at institutions including CEA Saclay, Institut d'Astrophysique de Paris, and others to study galaxy morphology, black hole growth, and environment-dependent processes.

Overview

Horizon-AGN was executed to reproduce observational constraints such as the stellar mass function, galaxy morphology demographics, and the cosmic star formation history while resolving processes from the scale of supermassive black hole accretion to the cosmic web filamentary network. The simulation volume (100 Mpc/h box) balances sample statistics with spatial resolution to investigate phenomena relevant to surveys like Sloan Digital Sky Survey, COSMOS, CANDELS, and facilities such as Hubble Space Telescope and James Webb Space Telescope. The collaboration builds on prior simulation efforts exemplified by projects including Illustris, EAGLE, Millennium Simulation, and Magneticum Pathfinder.

Simulation Design and Physics

The physical model incorporates a ΛCDM cosmology consistent with WMAP7 parameters and follows collisionless dark matter with hydrodynamics for baryons. Key subgrid physics modules include radiative cooling and heating with a uniform ultraviolet background motivated by Haardt & Madau, metallicity-dependent cooling tied to yields from Type Ia supernova, Type II supernova, and asymptotic giant branch enrichment, and a star formation prescription calibrated to the Kennicutt–Schmidt law. Stellar feedback implements kinetic and thermal energy injection informed by studies of supernova remnant energetics and chemical enrichment patterns observed in systems such as Milky Way and Andromeda. AGN feedback models incorporate both quasar-mode and radio-mode implementations to regulate star formation in massive galaxies, inspired by accretion physics explored in works about Bondi accretion and observational constraints from Chandra X-ray Observatory and ALMA.

Numerical Methods and Implementation

Horizon-AGN uses the adaptive mesh refinement (AMR) code RAMSES to solve the Euler equations and Poisson equation on a comoving grid, coupling particle-mesh techniques for N-body dynamics with finite-volume Godunov solvers for hydrodynamics. The simulation employs refinement criteria based on mass and density thresholds to achieve high spatial resolution in collapsed structures while preserving computational efficiency on supercomputing platforms such as GENCI-sponsored clusters and national facilities. Parallelization leverages Message Passing Interface practices common to codes like GADGET-2 and ENZO, and outputs are post-processed with halo finders such as AHF or friends-of-friends methods used in Rockstar and merger-tree builders comparable to Consistent Trees. Data products include catalogs of halos, subhalos, galaxy stellar properties, and black hole merger histories, enabling comparisons with datasets from GALEX, Spitzer Space Telescope, and WISE.

Key Results and Findings

Analyses from Horizon-AGN have shown that including AGN feedback is essential to reproduce the high-mass end cutoff of the stellar mass function and to quench star formation in massive ellipticals, paralleling inferences from studies of Brightest Cluster Galaxy populations and galaxy color bimodality observed in DEEP2 and VIPERS. The simulation reproduces trends in galaxy size evolution, morphological transformation driven by mergers and disk instabilities, and the co-evolution of galaxies and supermassive black holes consistent with empirical relations such as the M–sigma relation and the black hole mass–stellar mass relation. Horizon-AGN has been used to study cosmic magnetogenesis, gas accretion modes (cold versus hot flows), environmental influences on satellite quenching analogous to processes inferred in Virgo Cluster and Coma Cluster, and the imprint of feedback on the intergalactic medium and the Lyman-alpha forest.

Comparisons and Validation

The simulation has been systematically compared with other major projects, notably IllustrisTNG, EAGLE, SIMBA, and MassiveBlack-II, to evaluate differences arising from subgrid prescriptions, resolution, and hydrodynamical solvers. Validation efforts include matching the observed galaxy stellar mass function, luminosity functions across bands used by SDSS and CFHTLS, galaxy color distributions from CANDELS and COSMOS, black hole mass functions gleaned from AGN surveys such as SDSS Quasar Catalog and X-ray studies with XMM-Newton, and clustering statistics compared with measurements from BOSS and 2dF Galaxy Redshift Survey.

Applications and Impact

Horizon-AGN outputs underpin studies of galaxy formation theory, informing semi-analytic models like those implemented in GALFORM and empirical models such as UniverseMachine. The simulation has provided mock catalogs for planning and interpreting data from upcoming surveys including Euclid, LSST (Vera C. Rubin Observatory), and WFIRST (Nancy Grace Roman Space Telescope), and has contributed to work on gravitational lensing forecasts, feedback energetics relevant to Sunyaev–Zel'dovich effect studies, and predictions for multiwavelength observables spanning radio to X-ray regimes exploited by facilities like SKA and Chandra. Its publicly shared datasets and derived catalogs have influenced theoretical development across collaborations at institutions such as CEA, CNRS, Observatoire de Paris, and numerous university groups.

Category:Cosmological simulations Category:Galaxy formation