The Feedback In Realistic Environments (FIRE) project explores feedback in cosmological galaxy formation simulations. Previous FIRE simulations used an identical source code ("FIRE-1") for consistency. Motivated by the development of more accurate numerics -including hydrodynamic solvers, gravitational softening, and supernova coupling algorithms -and exploration of new physics (e.g. magnetic fields), we introduce "FIRE-2", an updated numerical implementation of FIRE physics for the GIZMO code. We run a suite of simulations and compare against FIRE-1: overall, FIRE-2 improvements do not qualitatively change galaxy-scale properties. We pursue an extensive study of numerics versus physics. Details of the star-formation algorithm, cooling physics, and chemistry have weak effects, provided that we include metal-line cooling and star formation occurs at higher-than-mean densities. We present new resolution criteria for high-resolution galaxy simulations. Most galaxy-scale properties are robust to numerics we test, provided: (1) Toomre masses are resolved; (2) feedback coupling ensures conservation, and (3) individual supernovae are time-resolved. Stellar masses and profiles are most robust to resolution, followed by metal abundances and morphologies, followed by properties of winds and circum-galactic media (CGM). Central (∼kpc) mass concentrations in massive (> L * ) galaxies are sensitive to numerics (via trapping/recycling of winds in hot halos). Multiple feedback mechanisms play key roles: supernovae regulate stellar masses/winds; stellar mass-loss fuels late star formation; radiative feedback suppresses accretion onto dwarfs and instantaneous star formation in disks. We provide all initial conditions and numerical algorithms used.

We present a suite of 15 cosmological zoom-in simulations of isolated dark matter halos, all with masses of M halo ≈ 10 10 M at z = 0, in order to understand the relationship between halo assembly, galaxy formation, and feedback's effects on the central density structure in dwarf galaxies. These simulations are part of the Feedback in Realistic Environments (FIRE) project and are performed at extremely high resolution (m baryon = 500 M , m dm = 2500 M ). The resultant galaxies have stellar masses that are consistent with rough abundance matching estimates, coinciding with the faintest galaxies that can be seen beyond the virial radius of the Milky Way (M /M ≈ 10 5 − 10 7 ). This non-negligible spread in stellar mass at z = 0 in halos within a narrow range of virial masses is strongly correlated with central halo density or maximum circular velocity V max , both of which are tightly linked to halo formation time. Much of this dependence of M on a second parameter (beyond M halo ) is a direct consequence of the M halo ∼ 10 10 M mass scale coinciding with the threshold for strong reionization suppression: the densest, earliest-forming halos remain above the UV-suppression scale throughout their histories while late-forming systems fall below the UV-suppression scale over longer periods and form fewer stars as a result. In fact, the latest-forming, lowest-concentration halo in our suite fails to form any stars. Halos that form galaxies with M 2 × 10 6 M have reduced central densities relative to dark-matter-only simulations, and the radial extent of the density modifications is well-approximated by the galaxy half-mass radius r 1/2 . Lower-mass galaxies do not modify their host dark matter halos at the mass scale studied here. This apparent stellar mass threshold of M ≈ 2 × 10 6 ≈ 2 × 10 −4 M halo is broadly consistent with previous work and provides a testable prediction of FIRE feedback models in ΛCDM. ‡ Caltech-Carnegie Fellow substantially more difficult and less conclusive. The difficulty is two-fold: these small scales are firmly in the non-linear regime of cosmological density perturbations at z = 0, meaning analytic approaches that are appropriate and straightforward for large scales no longer apply, and the galaxies that trace non-linear structure on small scales (dwarf galaxies) are inherently low-luminosity and small, making them difficult to study over cosmological scales.Over the past two decades, improvements in instrumentation and observations have provided dramatically improved data on the

We study a suite of extremely high-resolution cosmological FIRE simulations of dwarf galaxies (M halo 10 10 M ), run to z = 0 with 30 M resolution, sufficient (for the first time) to resolve the internal structure of individual supernovae remnants within the cooling radius. Every halo with M halo 10 8.6 M is populated by a resolved stellar galaxy, suggesting very low-mass dwarfs may be ubiquitous in the field. Our ultra-faint dwarfs (UFDs; M * < 10 5 M ) have their star formation truncated early (z 2), likely by reionization, while classical dwarfs (M * > 10 5 M ) continue forming stars to z < 0.5. The systems have bursty star formation (SF) histories, forming most of their stars in periods of elevated SF strongly clustered in both space and time. This allows our dwarf with M * /M halo > 10 −4 to form a dark matter core > 200 pc, while lower-mass UFDs exhibit cusps down to 100 pc, as expected from energetic arguments. Our dwarfs with M * > 10 4 M have half-mass radii (R 1/2 ) in agreement with Local Group (LG) dwarfs; dynamical mass vs. R 1/2 and the degree of rotational support also resemble observations. The lowest-mass UFDs are below surface brightness limits of current surveys but are potentially visible in next-generation surveys (e.g. LSST). The stellar metallicities are lower than in LG dwarfs; this may reflect pre-enrichment of the LG by the massive hosts or Pop-III stars. Consistency with lower resolution studies implies that our simulations are numerically robust (for a given physical model).

The vast majority of dwarf satellites orbiting the Milky Way and M31 are quenched, while comparable galaxies in the field are gas-rich and star-forming. Assuming that this dichotomy is driven by environmental quenching, we use the ELVIS suite of N -body simulations to constrain the characteristic timescale upon which satellites must quench following infall into the virial volumes of their hosts. The high satellite quenched fraction observed in the Local Group demands an extremely short quenching timescale (∼ 2 Gyr) for dwarf satellites in the mass range M ∼ 10 6 − 10 8 M . This quenching timescale is significantly shorter than that required to explain the quenched fraction of more massive satellites (∼ 8 Gyr), both in the Local Group and in more massive host halos, suggesting a dramatic change in the dominant satellite quenching mechanism at M 10 8 M . Combining our work with the results of complementary analyses in the literature, we conclude that the suppression of star formation in massive satellites (M ∼ 10 8 − 10 11 M ) is broadly consistent with being driven by starvation, such that the satellite quenching timescale corresponds to the cold gas depletion time. Below a critical stellar mass scale of ∼ 10 8 M , however, the required quenching times are much shorter than the expected cold gas depletion times. Instead, quenching must act on a timescale comparable to the dynamical time of the host halo. We posit that ram-pressure stripping can naturally explain this behavior, with the critical mass (of M ∼ 10 8 M ) corresponding to halos with gravitational restoring forces that are too weak to overcome the drag force encountered when moving through an extended, hot circumgalactic medium.