CAMELS

Warning

A bug was recently identified in the generation of the FOF/Subfind catalogs for SIMBA, SIMBA_DM and Illustris_DM 1P and CV sets, as well as a small number of Astrid_DM 1P simulations and EX simulations. The bug affected mostly global DM and gas properties of FOF/Subfind objects, less so their galaxy-scale properties such as their stellar properties. We believe that in particular some studies using these 1P sets might have been non-negligibly affected by this. A separate bug affected the FOF/Subfind catalogs for all sets (1P, CV, LH) of Swift-EAGLE. We welcome hearing from anyone who is concerned their study might have been affected, so that we can help figure things out. All the data in the public FOF/Subfind directories is now fixed, and the old buggy data are available upon request. Apologies for any inconvenience!

CAMELS stands for Cosmology and Astrophysics with MachinE Learning Simulations, and it is a project that aims at building bridges between cosmology and astrophysics through numerical simulations and machine learning.

As of March 2025, CAMELS host more than 2 petabytes of data from 16,960 cosmological simulations: 7,208 N-body and 9,752 hydrodynamic simulations. This project is the result of a large, collaborative effort and represents the most extensive suite of cosmological hydrodynamic simulations ever conducted.

Type

Code

Subgrid model

Simulations / Generation

First

Second

Third

Hydrodynamic

Arepo

IllustrisTNG

3,219

1,192

Gizmo

SIMBA

1,171

MP-Gadget

Astrid

2,080

OpenGadget

Magneticum

77

Swift

EAGLE

1,052

Ramses

552

48

Enzo

6

Gadget4-Osaka

CROCODILE

260

148

Gizmo

Obsidian

27

N-body

Gadget-III

6,136

1,072

Introductory video to the CAMELS project


The video below shows an example of a CAMELS hydrodynamic simulation run with the Ramses code. Gas density and gas temperature are shown in blue and red, respectively as a function of time. CAMELS contains thousands of simulations like this one.


The video below illustrates the differences between different hydrodynamic simulations. All simulations share the same cosmology, initial conditions, and have fiducial astrophysics values. Differences in the different fields shown are due to the intrinsic differences between the subgrid models employed in the simulatons.


The video shows a comparison between a first- and second-generation CAMELS simulation. Both simulations have the same mass and spatial resolution, but the second-generation one, has a volume 8x larger than its first-generation one.


Do you want to play a game?
See how well you can distinguish between different physical fields created from CAMELS.

CAMELS have AI agents to help you with data, paper, coding...etc. Click on the image to use them!
CAMELS Agent Logo

Other