Project summary

The FIRE (Feedback In Realistic Environments) project, introduced in Hopkins et al. (2014), seeks to improve the predictive power of simulations of galaxy formation, by developing cosmological simulations that resolve the interstellar medium of individual galaxies, while implementing the major channels of stellar feedback as directly as possible from stellar evolution models, all in realistic cosmological environments. FIRE simulations typically zoom in on a region around a single primary galaxy or group of galaxies, to achieve parsec-scale resolution within a cosmological context. For more details see the FIRE homepage.

This is the main page for accessing publicly available FIRE-2 simulation data. Most FIRE-2 data is in HDF5 file format.

Documentation

Wetzel et al. (2023) is the primary documentation (and primary citation) for FIRE-2 public data. It describes Data Release 1 (DR1), including explanations of the physical models that go into the FIRE-2 simulations, the contents of snapshots, the catalogs of (sub)halos and their galaxies, and various analysis tools

Wetzel et al. (2025) describes the new data in Data Release 2 (DR2), including simulations with physics variations, and halo merger trees

Hopkins et al. (2018) describes the FIRE-2 physics model used to generate these simulations

Hopkins 2015 and this website describe the publicly-available Gizmo code used to generate the FIRE simulations

This page provides more detailed documentation of the contents of Gizmo snapshots files

License and citing

We release FIRE-2 data under the Creative Commons BY 4.0 license. If you use these data, we request that you cite as follows:

“We use the publicly-available FIRE-2 cosmological zoom-in simulations (Wetzel et al. 2023, 2025), from the Feedback In Realistic Environments (FIRE) project, generated using the Gizmo code (Hopkins 2015) and the FIRE-2 physics model (Hopkins et al. 2018)."

We also request users to cite the individual published article(s) that introduced each simulation used, as listed in Tables 1, 2, 3 in Wetzel et al. (2023) for the simulations first released in Data Release 1, or as listed in Wetzel et al. (2025) for the simulations first released in Data Release 2. See also below.

Contents

Core suite to z = 0

The Core suite comprises 20 simulations that zoom in on 14 Milky Way-mass galaxies, 5 SMC/LMC-mass galaxies, and 4 lower-mass galaxies, including 1 ultrafaint. For each simulation, we include all 601 snapshots across z = 0 − 99, including catalogs of (sub)halos and their galaxies at all snapshots and full merger trees. We also include files for tracking particles across snapshots, and for all star particles today, their 3-D formation coordinates and an "ex-situ" flag. For the Milky Way-mass galaxies, we also include catalogs of streams, and various representations of the gravitational potential.

The following articles introduced each of these simulations. Anyone using a given simulation also should cite that article.

m09, m10q, m10v: Wheeler et al. (2019)

m11b: Chan et al. (2018)

m11d, m11e, m11h, m11i: El-Badry et al. (2018)

m11q, m12m: Hopkins et al. (2018)

m12i: Wetzel et al. (2016)

m12z, m12b, m12c: Garrison-Kimmel et al. (2019a)

m12f, m12_elvis_RomeoJuliet, m12_elvis_ThelmaLouise: Garrison-Kimmel et al. (2017)

m12_elvis_RomulusRemus: Garrison-Kimmel et al. (2019b)

m12r, m12w: Samuel et al. (2020)

For a subset of Core simulations to z = 0, we also include the following physics variations:

Dark Matter Only (19 simulations): model only dark matter (users should cite the work that introduced the corresponding baryonic simulation, as listed above and in Table 1 of Wetzel et al. 2023)

Later Reionization (4 simulations): use an updated ultraviolet background, which undergoes later cosmic reionization at z = 7.8, instead of at z ~ 10 as in other FIRE-2 simulations (introduced in/cite Gandhi et al. 2022 for m09, Wetzel et al. 2025 for m12f, m12i, m12m)

MHD+ (16 simulations): include magnetohydrodynamics, as well as anisotropic conduction and viscosity in gas (introduced in/cite Hopkins et al. 2020)

Cosmic Ray (14 simulations): in addition to MHD+, these model the injection, transport, and feedback of cosmic rays from supernovae, assuming a constant diffusion coefficient (introduced in/cite Hopkins et al. 2020)

Massive Halo suite to z = 1

The Massive Halo suite comprises 8 massive galaxies: A1, A2, A4, A8 from Anglés-Alcázar et al. (2017), and B1, B2, C1, C2 from Cochrane et al. (2023), all run to z = 1. We selected these halos from the FIRE-1 MassiveFIRE suite (Feldmann et al. 2016, 2017). For each simulation, we release 278 snapshots across z = 1 - 99. These halos have masses \(M_{\rm halo} = 10^{12-13.5} M_{\odot}\) at z = 1. Four simulations (A1, A2, A4, A8) include massive black hole growth, but none of these simulations include AGN feedback, which results in overly massive galaxies with ultra-dense nuclear stellar distributions at late times. Anyone using these simulations should cite Anglés-Alcázar et al. (2017) and/or Cochrane et al. (2023).

High Redshift suite to z = 5

This suite comprises 34 simulations that target halos with \(M_{\rm halo} = 10^{9-12} M_{\odot}\) at z = 5 - 9, from Ma et al. (2018), Ma et al. (2019), Ma et al. (2020). 22 of these simulations are run from z = 99 to z = 5 (with 68 snapshots), 6 are run to z = 7 (with 42 snaphots), and 6 are run to z = 9 (with 27 snasphots). In addition to the primary galaxy, there are many lower-mass galaxies within each zoom-in region, so the High Redshift suite contains thousands of resolved galaxies at each snapshot. Anyone using these simulations should the Ma et al. articles above.

Cosmological Boxes to z = 0

This suite comprises 4 dark-matter-only cosmological boxes that we used to generate zoom-in initial conditions for many (but not all) of the FIRE simulations. These boxes have comoving side lengths of 86 Mpc, 108 Mpc, 136 Mpc, and 172 Mpc, each each one differs in mass resolution by a factor of 2. Anyone using these cosmological boxes should cite Wetzel et al. (2025).