Project summary

The AGORA High-resolution Galaxy Simulations Comparison Project investigates galaxy formation with high-resolution simulations and compares the results across code platforms and with observations. The goal is to verify that the astrophysical assumptions are responsible for any success in galaxy simulations, not artifacts of particular numerical implementations. By guaranteeing reproducibility, we aim to improve numerical experiments as one of the most powerful tools in formulating theories of galaxy formation. This is a one-of-a-kind, inter-institutional effort by more than 160 participants from over 60 institutions worldwide. We welcome anybody who is interested in participating in the AGORA project, especially with new codes.

The AGORA infrastructure include common cosmological initial conditions (4 halo masses and 2 assembly histories generated using MUSIC), common isolated initial conditions (generated with MakeDisk), common astrophysics models (cooling and UV background implemented via the GRACKLE package), and a common analysis platform (yt toolkit), all publicly available to the community. Using the components assembled, the ongoing collaborative efforts aim to increase the realism of numerical experiments collectively, resulting in multiple publications in the next few years.

The data are available here.

For more information about the project, the simulation data, or the infrastructures assembled, please visit the AGORA homepage.

Any publication using these data should cite the relevant paper(s) linked below:

• Paper I - AGORA rationale and dark matter-only comparison: Kim et al. (2014)
• Paper II - isolated disk comparison: Kim et al. (2016)
• Paper II - public data release: Roca-Fabrega et al. (2020)
• Paper III - cosmological comparison: Roca-Fabrega et al. (2021)
• Paper IV - Cosmological comparison at Low redshift + merger trees: Roca-Fabrega et al. (2024)

The data from Paper II, III and IV are publicly available and can be found here. All the CosmoRun data saved in more than 300 fine time steps, will be available soon.

Paper IV data structure

The Paper IV Data Release note with full information about how to find and download the simulations snapshots and the metadata can be found here. In the released data, we include 14 snapshots from each code at the integer redshifts between 15 and 2, plus the ones at z=1 and 0, when available. You can find the snapshot number (included in the snapshot filename) corresponding to each specific redshift in the following table:

• i: ART-I
• ii: ENZO
• iii: RAMSES
• iv: CHANGA
• v: CHANGA-T
• vi: GADGET-3
• vii: GEAR
• viii: AREPO-T
• ix: GIZMO

Together with the snapshots data we also released several metadata files. These files are located in the metadata folder you can find here and contains the following three files and a subfolder:

• centers_CODE_final_noID.txt: This file contains the centers of the main halo inside the simulated cosmological box in physical kiloparsecs. The data is structured as follows: redshift, snapnum (see Table 2), [x,y,z].
• Rvir_cent_CODE.txt: This file contains similar information to the previous one but in a simpler format, including information on the R200c. The data is structured as follows: redshift, snapnum (see Table 2), x, y, z, R200c (in physical kpc).
• Rvir_cent_others_CODE.txt: This file contains information about many general properties of the central halo. The data is structured as follows: redshift, snapnum (see Table 2), R200c [kpc], M200c [M⊙], M⊙(R<R200c) [M⊙], M⊙,gal(R< 0.15R200c) [M⊙], R1/2 [kpc], Mtot(R<R1/2 ) [M⊙], Z⊙ [Z⊙], R1/2⊙,gal [kpc], M⊙,gal(R<R1/2⊙ ) [M⊙], Z⊙,gal [Z⊙].
• MERGERTREES_ROCKSTAR: In this folder you will find a sub-folder containing the merger for each one of the participant codes except for the fiducial CHANGA (CHANGA_3.5e51erg_SB), generated using rockstar (Behroozi et al. 2013a) and consistent-trees (Behroozi et al. 2013b). The files can be read and analysed using the ytree tool provided by the yt community.

All metadata filenames include the name of the corresponding code plus some small variations indicating its content.

WARNING: Because the CHANGA fiducial (CHANGA_3.5e51erg_SB) merger trees are not yet available, we provide the CHANGA-T (CHANGA_6.0e51erg_NSB) ones instead. Although this can give a first guess of the center of all subhalos and of the central halo, we recommend the users to make additional checks and to apply a re-centering routine before using the data (e.g. centering at the center of mass in a sphere centered at the center provided in the merger tree catalog).