FIREbox Cosmological Simulation

• FIREbox is a large-volume, high-dynamic-range cosmological simulation that models galaxy formation from first principles using comprehensive baryonic physics.
• It combines state-of-the-art resolution (~20 pc spatial, ~6×10⁴ M_⊙ baryonic elements) with a 22.1 cMpc volume to capture robust galaxy populations from dwarfs to Milky Way-mass systems.
• The simulation employs multi-channel stellar feedback, detailed cooling and chemical enrichment to investigate galaxy scaling relations, morphologies, and interstellar medium dynamics within a ΛCDM framework.

The FIREbox Cosmological Simulation is a large-volume, high-dynamic-range numerical experiment designed within the Feedback in Realistic Environments (FIRE) project to model galaxy formation and evolution from first principles. It is notable for combining the spatial resolution typical of state-of-the-art zoom-in simulations (~20 pc, baryonic mass elements of ~6 × 10⁴ M_⊙) with a representative cosmological volume (22.1 cMpc per side), allowing for statistically robust sampling of galaxy populations down to dwarf scales and capturing the full cosmological context of structure formation. FIREbox implements a comprehensive suite of baryonic physics—including multiphase ISM treatment, explicit multi-channel stellar feedback, and detailed modeling of star formation and chemical enrichment—directly calibrated from stellar evolution and population synthesis models, all embedded in a consistent ΛCDM cosmology. The simulation is used to probe galaxy scaling relations, morphologies, interstellar and circumgalactic gas properties, and the role of feedback and environment, revealing strengths and tensions relative to observations and advancing the theoretical landscape of galaxy formation.

1. Simulation Design, Physical Ingredients, and Resolution

FIREbox is a fully cosmological hydrodynamic simulation with an effective dynamic range of ~10⁶, employing the GIZMO code with the FIRE-2 physics suite (Feldmann et al., 2022). The volume is a periodic cube of L = 22.1 cMpc and resolves ~1000 galaxies above 10⁸ M_⊙ in stellar mass at z = 0. The simulation achieves a mass resolution of m_b ≈ 6 × 10⁴ M_⊙ for baryons and a spatial resolution in dense gas of ~20 pc. The high resolution ensures proper sampling of the multiphase ISM and accurate capturing of individual giant molecular clouds, critical for realistic star formation and feedback processes.

The model includes:

• Detailed cooling physics (atomic, molecular, and metal-line) spanning 10 K–10¹⁰ K
• Star formation occurring only in dense, self-gravitating, molecular regions
• Multi-channel feedback: supernovae (Type II/Ia), stellar winds, radiation pressure, photoionization and photoelectric heating
• Metal enrichment and passive scalars for chemical tagging

Initial conditions are drawn to match Planck ΛCDM cosmological parameters. All major baryonic processes—cooling, star formation, and supernova feedback—are “physics-derived” without explicit parameter tuning for outcome matching.

2. Galaxy Properties: Scaling Relations and Comparisons to Observations

FIREbox galaxies below M_star ≈ 10¹⁰.⁵–10¹¹ M_⊙ robustly match observed galaxy scaling relations:

• The simulated star-forming main sequence shows a near-linear relation between SFR and M_star (in log-log space) at z ≈ 0–3, consistent with observations.
• Gas fractions (atomic and molecular) and mass–metallicity relations follow observed (often broken power-law) scaling laws with stellar mass, exhibiting characteristic bends and flattening at high masses (Feldmann et al., 2022).
• The baryonic Tully–Fisher and stellar mass–halo mass relations are broadly reproduced for intermediate-mass galaxies (Feldmann et al., 2022, Ma et al., 2017).

However, at M_star ≳ 10¹¹ M_⊙, FIREbox predicts galaxy abundances and cosmic SFR densities exceeding observational estimates, indicating that stellar feedback alone is insufficient for quenching the most massive galaxies within the constraints of standard FIRE-2 physics.

The galaxy stellar mass function and various luminosity functions (rest-UV/optical) are in broad agreement with data from z ≈ 0 to z ≈ 12. In the high-redshift regime (z ≳ 6), FIREbox-HR (the high-resolution resimulation) matches the observed UV luminosity function and cosmic UV luminosity density, reproducing the relatively high star-formation activity inferred from early JWST data (Feldmann et al., 2 Jul 2024).

3. Star Formation, Feedback Regulation, and Gas Cycle

Star formation in FIREbox is self-regulated by explicit stellar feedback. The local recipe converts cold, dense, molecular gas into stars per local free-fall time (t_ff), i.e.,

$$\dot{\rho}_\star = \frac{\rho_\mathrm{mol}}{t_\mathrm{ff}}, \quad t_\mathrm{ff} = \sqrt{\frac{3\pi}{32\,G\,\rho}}$$

but the global, time-averaged efficiency is reduced to a few percent by feedback-driven turbulence and outflows (Hopkins et al., 2013).

Stellar feedback is implemented in multiple interacting channels:

• Direct injection of energy, momentum, mass, and metals from supernovae and stellar winds, with the energy/momentum derived from population synthesis models (e.g., STARBURST99). If the cooling radius of a supernova is unresolved, the final momentum is boosted analytically to capture unresolved Sedov–Taylor phase physics.
• Radiation pressure momentum deposition is computed as

$$\dot{p}_\mathrm{rad} \simeq \left(1 - e^{-\tau_{\rm UV/opt}}\right) (1 + \tau_{\rm IR}) \frac{L_{\rm incident}}{c}$$

with opacities scaling with gas metallicity.

• Photoionization and photoelectric heating are included by propagating the stellar ionizing luminosity to exhaustion in each star’s surroundings.

Feedback generates strong burst-to-burst SFR variability—dwarfs frequently experience order-of-magnitude changes in SFR on dynamical timescales—and sustains a turbulent, multiphase ISM. Global star formation histories are “flattened” or steadily rising relative to halo accretion rates, in contrast to sub-grid wind models which tend to produce peaked early star formation followed by artificial declines.

A significant fraction of feedback-driven outflows do not permanently escape the host halo but instead populate a metal-enriched reservoir in the CGM, which recycles onto the galaxy and maintains star formation over cosmological timescales (Muratov et al., 2015).

4. Morphology and Structural Properties

FIREbox galaxies display a mass-dependent morphology transition (Benavides et al., 1 Aug 2025, Klein et al., 7 Mar 2025):

• At $M_\star \gtrsim 10^{10}\ M_\odot$, a large fraction of galaxies have thin, rotationally supported stellar disks. The simulated axis ratios ($q\sim 0.1$) and Sersic-profile morphologies in mock r-band images are comparable to the thinnest disks in surveys such as GAMA.
• At $M_\star < 10^{10}\ M_\odot$, FIREbox produces a deficit of thin disks. The fraction of galaxies with projected $q<0.4$ drops rapidly, and no systems with $q<0.2$ are found. The transition regime ($10^{9} < M_\star/M_\odot < 10^{10}$) is characterized by decreasing burstiness in star formation, deeper host halo potentials, and the emergence of rotational support.
• Low-mass ($M_\star < 10^9\ M_\odot$) galaxies are predominantly spheroidal or elongated in their young and luminous stellar components, indicating that feedback-induced turbulence and shallow gravitational wells inhibit the formation of stable disks.

This result, although consistent with a trend toward thicker disks in observed dwarfs, is in tension with the existence of some observed rotationally supported low-mass disks, suggesting feedback models may over-disrupt disk formation in low-mass halos or that additional physics (e.g., improved CGM modeling, resolution) are required.

5. Cold Gas Structure and Outflows

FIREbox Milky Way–mass galaxies form cold HI disks with vertical scale heights ranging from ~100 pc in the center to ~800–1000 pc at large radii (Gensior et al., 2022). Five independent methods (including Gaussian fitting, half-width at half-maximum, hydrostatic equilibrium, and mass fraction cuts) yield consistent results for $h_\mathrm{HI}$, in excellent agreement with observed HI scale heights. The simulations demonstrate that achieving thin disks at this resolution is contingent not on numerical resolution alone, but on the interplay of explicitly modeled multiphase ISM cooling (to 10 K) and feedback channels that maintain realistic turbulence and prevent unphysical heating of the ISM.

Superbubble-driven, multiphase outflows, resolved in detail via high-resolution zoom-ins, exhibit:

• Burst-triggered peaks in outflow mass and energy, especially during clustered supernova activity, with typical hot phase breakout scales of several kpc.
• Cold outflow mass-loading factors and momentum fluxes in broad agreement with observational inferences and analytic wind theory, but energy-loading factors in some bursts that exceed predictions of tall–box simulations, partially due to burstier star formation histories and measurement choices (Porter et al., 5 Jun 2024).

6. Quenching, Post-starburst Phenomena, and Feedback Limits

FIREbox provides nuanced predictions for rapid star formation shut-off and the nature of post-starburst galaxies (Cenci et al., 29 Aug 2025):

• Simulated post-starburst (PSB) populations (selected from rest-frame color criteria) have a comparable fraction to observations (~10%), but only a minority (~8%) are truly quenched (“q-ASBs”) with sharply suppressed SFR and depleted molecular gas reservoirs.
• The majority are “impostors”: their PSB-like photometric signatures derive from effects other than genuine short-term quenching (e.g., stochastic star formation or viewing angle effects) and they retain active star formation and molecular gas.
• FIREbox simulations lacking black hole feedback under-produce the permanent cessation of star formation found in observations—implying that >70% of true PSBs in nature are likely attributable to massive black hole/AGN feedback rather than pure stellar feedback.
• Quenching mechanisms and timescales thus provide strong constraints on missing feedback physics and the co-evolution of galaxies and supermassive black holes.

7. Public Data Releases and Future Prospects

FIREbox and related FIRE-2 zoom-in simulations have extensive public data releases, currently in the second major release (Wetzel et al., 8 Aug 2025). Available products include:

• All snapshots for 14 MW-mass, 5 SMC/LMC-mass, and 4 lower-mass galaxies (Core suite; z ≈ 99–0)
• Massive Halo suite (z ≈ 99–1), high-redshift suite (z ≈ 99–5/7/9), with up to 601 snapshots per simulation and 25 Myr time spacing
• Physics variations: dark matter only, late reionization, inclusion of MHD+conduction/viscosity, and cosmic ray feedback
• Detailed catalogs of halos, subhalos, baryonic properties, merger trees, and gravitational potential expansions

Access is provided via the FlatHUB portal (http://flathub.flatironinstitute.org/fire), and data are released under a CC BY 4.0 license. The breadth and depth of the archive allow for diverse analyses, including galaxy assembly, satellite dynamics, feedback benchmarking, multi-wavelength mock observations, and model comparison.

Future work will focus on:

• Incorporating additional feedback sources (AGN, cosmic rays, magnetic fields)
• Finer mass and spatial resolution (e.g., FIREbox-HR: m_b ≈ 7.8 × 10³ M_⊙)
• Expanding theoretical models to resolve current tensions in dwarf morphology, disk formation, and star formation quenching

Summary Table: Select FIREbox Features and Findings

Aspect	Quantitative Scope / Result	Context / Implication
Cosmological Volume	22.1 cMpc cubic box, dynamic range ~10⁶	Resolves both MW-mass and dwarf galaxies in a statistically significant sample
Mass/Spatial Resolution	m_b ≈ 6×10⁴ M_⊙, ≈ 20 pc in dense gas (FIREbox); ≈ 7.8×10³ M_⊙ (FIREbox-HR)	Captures GMCs, enables direct ISM and feedback modeling
Morphology–Stellar Mass Relation	Disk-dominated for M_★ > 10¹⁰ M_⊙; deficit of disks below 10¹⁰ M_⊙ (Klein et al., 7 Mar 2025, Benavides et al., 1 Aug 2025)	Feedback/burstiness and potential depth control disk formation; tension with observed dwarfs
Star Formation Self-regulation	Galaxy-wide SFE ≈ few percent; regulated by turbulent, multi-channel feedback	Matches observed KS-law globally; maintains cosmic SFR at observed levels (for M_★ < 10¹¹ M_⊙)
HI Scale Height	h_HI ~ 100 pc (center) to 800–1000 pc (edge) (Gensior et al., 2022)	Excellent agreement with observed MW-mass spiral galaxies
Post-starburst Galaxies	Photometrically selected PSBs ≈ 10% of population; ~8% are "true" quenched (Cenci et al., 29 Aug 2025)	Stellar feedback alone insufficient for full PSB quenching; AGN feedback likely required

FIREbox thus serves as both a baseline for ΛCDM galaxy formation with explicitly modeled feedback and a diagnostic tool for identifying the physical drivers of discrepancies in unresolved regimes, offering constraints for next-generation simulations and theoretical models.