SearcharxivSearch

arXiv subjects

Zachary Hafen

Publications and source records attributed to Zachary Hafen.

At least 19 recordsLinked to original sources

Second public data release of the FIRE-2 cosmological zoom-in simulations of galaxy formation

We describe the second data release (DR2) of the FIRE-2 cosmological zoom-in simulations of galaxy formation, from the Feedback In Realistic Environments (FIRE) project, available at http://flathub.flatironinstitute.org/fire. DR2 includes all snapshots for most simulations, starting at z ~ 99, with all snapshot time spacings <~ 25 Myr. The Core suite -- comprising 14 Milky Way-mass galaxies, 5 SMC/LMC-mass galaxies, and 4 lower-mass galaxies -- includes 601 snapshots to z = 0. For the Core suite, we also release resimulations with physics variations: (1) dark-matter-only versions; (2) a modified ultraviolet background with later reionization at z = 7.8; (3) magnetohydrodynamics, anisotropic conduction, and viscosity in gas; and (4) a model for cosmic-ray injection, transport, and feedback (assuming a constant diffusion coefficient). The Massive Halo suite now includes 8 massive galaxies with 278 snapshots to z = 1. The High Redshift suite includes 34 simulations: in addition to the 22 simulations run to z = 5, we now include 12 additional simulations run to z = 7 and z = 9. We also release 4 dark-matter-only cosmological boxes used to generate zoom-in initial conditions for many FIRE simulations. Most simulations include catalogs of (sub)halos and galaxies at all available snapshots, and most Core simulations to z = 0 include full halo merger trees.

astro-ph.GA

Density, asymmetry and citation dynamics in scientific literature

Scientific behavior is often characterized by a tension between building upon established knowledge and introducing novel ideas. Here, we investigate whether this tension is reflected in the relationship between the similarity of a scientific paper to previous research and its eventual citation rate. To operationalize similarity to previous research, we introduce two complementary metrics to characterize the local geometry of a publication's semantic neighborhood: (1) \emph{density} ($ρ$), defined as the ratio between a fixed number of previously-published papers and the minimum distance enclosing those papers in a semantic embedding space, and (2) asymmetry ($α$), defined as the average directional difference between a paper and its nearest neighbors. We tested the predictive relationship between these two metrics and its subsequent citation rate using a Bayesian hierarchical regression approach, surveying $\sim 53,000$ publications across nine academic disciplines and five different document embeddings. While the individual effects of $ρ$ on citation count are small and variable, incorporating density-based predictors consistently improves out-of-sample prediction when added to baseline models. These results suggest that the density of a paper's surrounding scientific literature may carry modest but informative signals about its eventual impact. Meanwhile, we find no evidence that publication asymmetry improves model predictions of citation rates. Our work provides a scalable framework for linking document embeddings to scientometric outcomes and highlights new questions regarding the role that semantic similarity plays in shaping the dynamics of scientific reward.

cs.DL

Accretion onto disk galaxies via hot and rotating CGM inflows

Observed accretion rates onto the Milky-Way and other local spirals fall short of that required to sustain star formation for cosmological timescales. A potential avenue for this unseen accretion is an inflow in the volume-filling hot phase ($\sim10^6$ K) of the circumgalactic medium (CGM), as suggested by some cosmological simulations. Using hydrodynamic simulations and a new analytic solution valid in the slow-rotation limit, we show that a hot inflow spins up as it approaches the galaxy, while remaining hot, subsonic and quasi-spherical. At the radius of angular momentum support ($\approx15$ kpc for the Milky-Way) the hot flow flattens into a disk geometry and then cools from $\sim10^6$ K to $\sim10^4$ K at the disk-halo interface. Cooling affects all hot gas, rather than just a subset of individual gas clouds, implying that accretion via hot inflows does not rely on local thermal instability in contrast with 'precipitation' models for galaxy accretion. Prior to cooling and accretion the inflow completes $\sim t_{\rm cool}/t_{\rm ff}$ radians of rotation, where $t_{\rm cool}/t_{\rm ff}$ is the cooling time to free-fall time ratio in hot gas immediately outside the galaxy. The ratio $t_{\rm cool}/t_{\rm ff}$ may thus govern the development of turbulence and enhancement of magnetic fields in gas accreting onto low-redshift spirals. We argue that accretion via hot inflows can explain the observed truncation of nearby thin stellar disks at $\approx4$ disk radii. We also show that if rotating hot inflows are common in Milky-Way size disk galaxies, as predicted, then signatures should be observable with X-ray telescopes and FRB surveys.

astro-ph.GA

Great Balls of FIRE III: Modeling Black Hole Mergers from Massive Star Clusters in Simulations of Galaxies

After the nearly hundred gravitational-wave detections reported by the LIGO-Virgo-KAGRA Collaboration, the question of the cosmological origin of merging binary black holes (BBHs) remains open. The two main formation channels generally considered are from isolated field binaries or via dynamical assembly in dense star clusters. Here, we focus on understanding the dynamical formation of merging BBHs within massive clusters in galaxies of different masses. To this end, we apply a new framework to consistently model the formation and evolution of massive star clusters in zoom-in cosmological simulations of galaxies. Each simulation, taken from the FIRE project, provides a realistic star formation environment with a unique star formation history and hosts realistic giant molecular clouds that constitute the birthplace of star clusters. Combined with the code for star cluster evolution CMC, we are able to produce populations of dynamically formed merging BBHs across cosmic time in different environments. As the most massive star clusters preferentially form in dense massive clouds of gas, we find that, despite their low metallicities favourable to the creation of black holes, low-mass galaxies contain few massive clusters and therefore have a limited contribution to the global production of dynamically formed merging BBHs. Furthermore, we find that massive clusters can host hierarchical BBH mergers with clear identifiable physical properties. Looking at the evolution of the BBH merger rate in different galaxies, we find strong correlations between BBH mergers and the most extreme episodes of star formation. Finally, we discuss the implications for future LIGO-Virgo-KAGRA gravitational wave observations.

astro-ph.GA

What Causes The Formation of Disks and End of Bursty Star Formation?

As they grow, galaxies can transition from irregular/spheroidal with 'bursty' star formation histories (SFHs), to disky with smooth SFHs. But even in simulations, the direct physical cause of such transitions remains unclear. We therefore explore this in a large suite of numerical experiments re-running portions of cosmological simulations with widely varied physics, further validated with existing FIRE simulations. We show that gas supply, cooling/thermodynamics, star formation model, Toomre scale, galaxy dynamical times, and feedback properties do not have a direct causal effect on these transitions. Rather, both the formation of disks and cessation of bursty star formation are driven by the gravitational potential, but in different ways. Disk formation is promoted when the mass profile becomes sufficiently centrally-concentrated in shape (relative to circularization radii): we show that this provides a well-defined dynamical center, ceases to support the global 'breathing modes' which can persist indefinitely in less-concentrated profiles and efficiently destroy disks, promotes orbit mixing to form a coherent angular momentum, and stabilizes the disk. Smooth SF is promoted by the potential or escape velocity (not circular velocity) becoming sufficiently large at the radii of star formation that cool, mass-loaded (momentum-conserving) outflows are trapped/confined near the galaxy, as opposed to escaping after bursts. We discuss the detailed physics, how these conditions arise in cosmological contexts, their relation to other correlated phenomena (e.g. inner halo virialization, vertical disk 'settling'), and observations.

astro-ph.GA

A unified model for the co-evolution of galaxies and their circumgalactic medium: the relative roles of turbulence and atomic cooling physics

The circumgalactic medium (CGM) plays a pivotal role in regulating gas flows around galaxies and thus shapes their evolution. However, the details of how galaxies and their CGM co-evolve remain poorly understood. We present a new time-dependent two-zone model that self-consistently tracks not just mass and metal flows between galaxies and their CGM but also the evolution of the global thermal and turbulent kinetic energy of the CGM. Our model accounts for heating and turbulence driven by both supernova winds and cosmic accretion as well as radiative cooling, turbulence dissipation, and halo outflows due to CGM overpressurization. We demonstrate that, depending on parameters, the CGM can undergo a phase transition (``thermalization'') from a cool, turbulence-supported phase to a virial-temperature, thermally-supported phase. This CGM phase transition is largely determined by the ability of radiative cooling to balance heating from supernova winds and turbulence dissipation. We perform an initial calibration of our model to the FIRE-2 cosmological hydrodynamical simulations and show that it can approximately reproduce the baryon cycles of the simulated halos. In particular, we find that, for these parameters, the phase transition occurs at high-redshift in ultrafaint progenitors and at low redshift in classical $M_{\rm vir}\sim10^{11}M_{\odot}$ dwarfs, while Milky Way-mass halos undergo the transition at $z\approx0.5$. We see a similar transition in the simulations though it is more gradual, likely reflecting radial dependence and multi-phase gas not captured by our model. We discuss these and other limitations of the model and possible future extensions.

astro-ph.GA

Born this way: thin disc, thick disc, and isotropic spheroid formation in FIRE-2 Milky-Way-mass galaxy simulations

We investigate the formation of Milky-Way-mass galaxies using FIRE-2 LCDM cosmological zoom-in simulations by studying the orbital evolution of stars formed in the main progenitor of the galaxy, from birth to the present day. We classify in situ stars as isotropic spheroid, thick-disc, and thin-disc according to their orbital circularities and show that these components are assembled in a time-ordered sequence from early to late times, respectively. All simulated galaxies experience an early phase of bursty star formation that transitions to a late-time steady phase. This transition coincides with the time that the inner CGM virializes. During the early bursty phase, galaxies have irregular morphologies and new stars are born on radial orbits; these stars evolve into an isotropic spheroidal population today. The bulk of thick-disc stars form at intermediate times, during a clumpy-disc ``spin-up'' phase, slightly later than the peak of spheroid formation. At late times, once the CGM virializes and star formation ``cools down," stars are born on circular orbits within a narrow plane. Those stars mostly inhabit thin discs today. Broadly speaking, stars with disc-like or spheroid-like orbits today were born that way. Mergers onto discs and secular processes do affect kinematics in our simulations, but play only secondary roles in populating thick-disc and in situ spheroid populations at z=0. The age distributions of spheroid, thick disc, and thin disc populations scale self-similarly with the steady-phase transition time, which suggests that morphological age dating can be linked to the CGM virialization time in galaxies.

astro-ph.GA

On the Metallicities and Kinematics of the Circumgalactic Media of Damped Ly$α$ Systems at $z \sim 2.5$

We use medium- and high-resolution spectroscopy of close pairs of quasars to analyze the circumgalactic medium (CGM) surrounding 32 damped Ly$α$ absorption systems (DLAs). The primary quasar sightline in each pair probes an intervening DLA in the redshift range $1.6 10^{13}~\rm cm^{-2}$ within 100 kpc of DLAs are larger by $2σ$ than those measured in the CGM of Lyman break galaxies (C$_f(N_\text{CII})>0.89$ and C$_f(N_\text{SiII})=0.75_{-0.17}^{+0.12}$). Metallicity constraints derived from ionic ratios for nine CGM systems with negligible ionization corrections and $N_\text{HI}>10^{18.5}~\rm cm^{-2}$ show a significant degree of scatter (with metallicities/limits across the range $-2.06\lesssim\log Z/Z_{\odot}\lesssim-0.75$), suggesting inhomogeneity in the metal distribution in these environments. Velocity widths of $\text{CIV}\lambda1548$ and low-ionization metal species in the DLA vs. CGM sightlines are strongly ($>2σ$) correlated, suggesting they trace the potential well of the host halo over $R_\bot\lesssim300$ kpc scales. At the same time, velocity centroids for $\text{CIV}\lambda1548$ differ in DLA vs. CGM sightlines by $>100~\rm km\ s^{-1}$ for $\sim50\%$ of velocity components, but few components have velocities that would exceed the escape velocity assuming dark matter host halos of $\ge10^{12}M_\odot$.

astro-ph.GA

The Halo21 Absorption Modeling Challenge: Lessons From "Observing" Synthetic Circumgalactic Absorption Spectra

In the Halo21 absorption modeling challenge we generated synthetic absorption spectra of the circumgalactic medium (CGM), and attempted to estimate the metallicity, temperature, and density (Z, T, and nH) of the underlying gas using observational methods. We iteratively generated and analyzed three increasingly-complex data samples: ion column densities of isolated uniform clouds, mock spectra of 1--3 uniform clouds, and mock spectra of high-resolution turbulent mixing zones. We found that the observational estimates were accurate for both uniform cloud samples, with Z, T, and nH retrieved within 0.1 dex of the source value for >90% of absorption systems. In the turbulent-mixing scenario, the mass, temperature, and metallicity of the strongest absorption components were also retrieved with high accuracy. However, the underlying properties of the subdominant components were poorly constrained because the corresponding simulated gas contributed only weakly to the H I absorption profiles. On the other hand, including additional components beyond the dominant ones did improve the fit, consistent with the true existence of complex cloud structures in the source data.

astro-ph.GA

Public data release of the FIRE-2 cosmological zoom-in simulations of galaxy formation

We describe a public data release of the FIRE-2 cosmological zoom-in simulations of galaxy formation, available at http://flathub.flatironinstitute.org/fire, from the Feedback In Realistic Environments (FIRE) project. FIRE-2 simulations achieve parsec-scale resolution to explicitly model the multi-phase interstellar medium while implementing direct models for stellar evolution and feedback, including stellar winds, core-collapse and Ia supernovae, radiation pressure, photoionization, and photoelectric heating. We release complete snapshots from 3 suites of simulations. The first comprises 20 simulations that zoom in on 14 Milky Way-mass galaxies, 5 SMC/LMC-mass galaxies, and 4 lower-mass galaxies including 1 ultra-faint; we release 39 snapshots across z = 0 - 10. The second comprises 4 massive galaxies, with 19 snapshots across z = 1 - 10. Finally, a high-redshift suite comprises 22 simulations, with 11 snapshots across z = 5 - 10. Each simulation also includes dozens of resolved lower-mass (satellite) galaxies in its zoom-in region. Snapshots include all stored properties for all dark matter, gas, and star particles, including 11 elemental abundances for stars and gas, and formation times (ages) of star particles. We also release accompanying (sub)halo catalogs, which include galaxy properties and member star particles. For the simulations to z = 0, including all Milky Way-mass galaxies, we release the formation coordinates and an "ex-situ" flag for all star particles, pointers to track particles across snapshots, catalogs of stellar streams, and multipole basis expansions for the halo mass distributions. We describe publicly available python packages for reading and analyzing these simulations.

astro-ph.GA

Great Balls of FIRE II: The evolution and destruction of star clusters across cosmic time in a Milky Way-mass galaxy

The current generation of galaxy simulations can resolve individual giant molecular clouds, the progenitors of dense star clusters. But the evolutionary fate of these young massive clusters, and whether they can become the old globular clusters (GCs) observed in many galaxies, is determined by a complex interplay of internal dynamical processes and external galactic effects. We present the first star-by-star $N$-body models of massive ($N\sim10^5-10^7$) star clusters formed in a FIRE-2 MHD simulation of a Milky Way-mass galaxy, with the relevant initial conditions and tidal forces extracted from the cosmological simulation. We select 895 ($\sim 30\%$) of the YMCs with $ > 6\times10^4M_{\odot}$ from Grudić et al.~2022 and integrate them to $z=0$ using the Cluster Monte Carlo Code, \texttt{CMC}. This procedure predicts a MW-like system with 148 GCs, predominantly formed during the early, bursty mode of star formation. Our GCs are younger, less massive, and more core-collapsed than clusters in the Milky Way or M31. This results from the assembly history and age-metallicity relationship of the host galaxy: younger clusters are preferentially born in stronger tidal fields and initially retain fewer stellar-mass black holes, causing them to lose mass faster and reach core collapse sooner than older GCs. Our results suggest that the masses and core/half-light radii of GCs are shaped not only by internal dynamical processes, but also by the specific evolutionary history of their host galaxies. These results emphasize that $N$-body studies with realistic stellar physics are crucial to understanding the evolution and present-day properties of GC systems.

astro-ph.GA

The imprint of bursty star formation on alpha-element abundance patterns in Milky Way-like galaxies

Milky Way-mass galaxies in the FIRE-2 simulations demonstrate two main modes of star formation. At high redshifts star formation occurs in a series of short and intense bursts, while at low redshifts star formation proceeds at a steady rate with a transition from one mode to another at times ranging from 3 to 7 Gyr ago for different galaxies. We analyse how the mode of star formation affects iron and alpha-element abundance. We find that the early bursty regime imprints a measurable pattern in stellar elemental abundances in the form of a "sideways chevron" shape on the [Fe/H] - [O/Fe] plane and the scatter in [O/Fe] at a given stellar age is higher than when a galaxy is in the steady regime. That suggests that the evolution of [O/Fe] scatter with age provides an estimate of the end of the bursty phase. We investigate the feasibility of observing of this effect by adding mock observational errors to a simulated stellar survey and find that the transition between the bursty and steady phase should be detectable in the Milky Way, although larger observational uncertainties make the transition shallower. We apply our method to observations of the Milky Way from the Second APOKASC Catalog and estimate that the transition to steady star formation in the Milky Way happened 7-8 Gyrs ago, earlier than transition times measured in the simulations.

astro-ph.GA

Great Balls of FIRE I: The formation of star clusters across cosmic time in a Milky Way-mass galaxy

The properties of young star clusters formed within a galaxy are thought to vary in different interstellar medium (ISM) conditions, but the details of this mapping from galactic to cluster scales are poorly understood due to the large dynamic range involved in galaxy and star cluster formation. We introduce a new method for modeling cluster formation in galaxy simulations: mapping giant molecular clouds (GMCs) formed self-consistently in a FIRE-2 MHD galaxy simulation onto a cluster population according to a GMC-scale cluster formation model calibrated to higher-resolution simulations, obtaining detailed properties of the galaxy's star clusters in mass, metallicity, space, and time. We find $\sim 10\%$ of all stars formed in the galaxy originate in gravitationally-bound clusters overall, and this fraction increases in regions with elevated $Σ_{\rm gas}$ and $Σ_{\rm SFR}$, because such regions host denser GMCs with higher star formation efficiency. These quantities vary systematically over the history of the galaxy, driving variations in cluster formation. The mass function of bound clusters varies -- no single Schechter-like or power-law distribution applies at all times. In the most extreme episodes, clusters as massive as $7\times 10^6 M_\odot$ form in massive, dense clouds with high star formation efficiency. The initial mass-radius relation of young star clusters is consistent with an environmentally-dependent 3D density that increases with $Σ_{\rm gas}$ and $Σ_{\rm SFR}$. The model does not reproduce the age and metallicity statistics of old ($>11\rm Gyr$) globular clusters found in the Milky Way, possibly because it forms stars more slowly at $z>3$.

astro-ph.GA

Hot-mode accretion and the physics of thin-disk galaxy formation

We use FIRE simulations to study disk formation in z~0, Milky Way-mass galaxies, and conclude that a key ingredient for the formation of thin stellar disks is the ability for accreting gas to develop an aligned angular momentum distribution via internal cancellation *prior* to joining the galaxy. Among galaxies with a high fraction (>70%) of their young stars in a thin disk (h/R~0.1) we find that: (i) hot, virial-temperature gas dominates the inflowing gas mass on halo scales (>~20 kpc), with radiative losses offset by compression heating; (ii) this hot accretion proceeds until angular momentum support slows inward motion, at which point the gas cools to T~10^4 K or less; (iii) prior to cooling, the accreting gas develops an angular momentum distribution that is aligned with the galaxy disk, and while cooling transitions from a quasi-spherical spatial configuration to a more flattened, disk-like configuration. We show that the existence of this "rotating cooling flow" accretion mode is strongly correlated with the fraction of stars forming in a thin disk among a sample of 17 z~0 galaxies spanning a halo mass range of 10^10.5 solar masses to 10^12 solar masses, or a stellar mass range 10^8 solar masses to 10^11 solar masses. Notably, galaxies with a thick disk or irregular morphology do not undergo significant angular momentum alignment of gas prior to accretion and show no correspondence between halo gas cooling and flattening. Our results suggest that rotating cooling flows (or, more generally, rotating subsonic flows) that become coherent and angular momentum-supported prior to accretion onto the galaxy are likely a necessary condition for the formation of thin, star-forming disk galaxies in a LambdaCDM universe.

astro-ph.GA

Amplified J-factors in the Galactic Center for velocity-dependent darkmatter annihilation in FIRE simulations

We use FIRE-2 zoom cosmological simulations of Milky Way size galaxy halos to calculate astrophysical J-factors for dark matter annihilation and indirect detection studies. In addition to velocity-independent (s-wave) annihilation cross sections $σ_v$, we also calculate effective J-factors for velocity-dependent models, where the annihilation cross section is either either p-wave ($\propto v^2/c^2$) or d-wave ($\propto v^4/c^4$). We use 12 pairs of simulations, each run with dark-matter-only (DMO) physics and FIRE-2 physics. We observe FIRE runs produce central dark matter velocity dispersions that are systematically larger than in DMO runs by factors of $\sim 2.5-4$. They also have a larger range of central ($\sim 400$ pc) dark matter densities than the DMO runs ($ρ_{\rm FIRE}/ρ_{\rm DMO} \simeq 0.5 - 3$) owing to the competing effects of baryonic contraction and feedback. At 3 degrees from the Galactic Center, FIRE J-factors are $5-50$ (p-wave) and $15-500$ (d-wave) times higher than in the DMO runs. The change in s-wave signal at 3 degrees is more modest and can be higher or lower ($\sim 0.3-6$), though the shape of the emission profile is flatter (less peaked towards the Galactic Center) and more circular on the sky in FIRE runs. Our results for s-wave are broadly consistent with the range of assumptions in most indirect detection studies. We observe p-wave J-factors that are significantly enhanced compared to most past estimates. We find that thermal models with p-wave annihilation may be within range of detection in the near future.

astro-ph.GA

Rapid disc settling and the transition from bursty to steady star formation in Milky Way-mass galaxies

Recent observations and simulations indicate substantial evolution in the properties of galaxies with time, wherein rotationally-supported and steady thin discs (like those frequently observed in the local universe) emerge from galaxies that are clumpy, irregular, and have bursty star formation rates (SFRs). To better understand the progenitors of local disc galaxies we carry out an analysis of three FIRE-2 simulated galaxies with a mass similar to the Milky Way at redshift z=0. We show that all three galaxies transition from bursty to steady SFRs at a redshift between z=0.5 and z=0.8, and that this transition coincides with a rapid (< ~1 Gyr) emergence of a rotationally-supported interstellar medium (ISM).In the late phase with steady SFR, the rotational energy comprises > ~90% of the total kinetic + thermal energy in the ISM, and is roughly half the gravitational energy. By contrast, during the early phase with bursty star formation, the ISM has a quasi-spheroidal morphology and its energy budget is dominated by quasi-isotropic flows including turbulence and coherent inflows/outflows. This result, that rotational support is subdominant at early times, challenges the common application of equilibrium disc models to the high-redshift progenitors of Milky Way-like galaxies. We further find that the formation of a rotation-supported ISM coincides with the formation of a thermal energy-supported inner circumgalactic medium (CGM). Before this transition, the inner CGM is also supported by turbulence and coherent flows, indicating that at early times there is no clear boundary between the ISM and inner CGM.

astro-ph.GA

Galaxies lacking dark matter produced by close encounters in a cosmological simulation

The standard cold dark matter plus cosmological constant model predicts that galaxies form within dark-matter haloes, and that low-mass galaxies are more dark-matter dominated than massive ones. The unexpected discovery of two low-mass galaxies lacking dark matter immediately provoked concerns about the standard cosmology and ignited explorations of alternatives, including self-interacting dark matter and modified gravity. Apprehension grew after several cosmological simulations using the conventional model failed to form adequate numerical analogues with comparable internal characteristics (stellar masses, sizes, velocity dispersions and morphologies). Here we show that the standard paradigm naturally produces galaxies lacking dark matter with internal characteristics in agreement with observations. Using a state-of-the-art cosmological simulation and a meticulous galaxy-identification technique, we find that extreme close encounters with massive neighbours can be responsible for this. We predict that approximately 30 percent of massive central galaxies (with at least 1e11 solar masses in stars) harbour at least one dark-matter-deficient satellite (with 1e8 - 1e9 solar masses in stars). This distinctive class of galaxies provides an additional layer in our understanding of the role of interactions in shaping galactic properties. Future observations surveying galaxies in the aforementioned regime will provide a crucial test of this scenario.

astro-ph.GA

Characterizing mass, momentum, energy and metal outflow rates of multi-phase galactic winds in the FIRE-2 cosmological simulations

We characterize mass, momentum, energy and metal outflow rates of multi-phase galactic winds in a suite of FIRE-2 cosmological "zoom-in" simulations from the Feedback in Realistic Environments (FIRE) project. We analyze simulations of low-mass dwarfs, intermediate-mass dwarfs, Milky Way-mass halos, and high-redshift massive halos. Consistent with previous work, we find that dwarfs eject about 100 times more gas from their interstellar medium (ISM) than they form in stars, while this mass "loading factor" drops below one in massive galaxies. Most of the mass is carried by the hot phase ($>10^5$ K) in massive halos and the warm phase ($10^3-10^5$ K) in dwarfs; cold outflows ($<10^3$ K) are negligible except in high-redshift dwarfs. Energy, momentum and metal loading factors from the ISM are of order unity in dwarfs and significantly lower in more massive halos. Hot outflows have $2-5\times$ higher specific energy than needed to escape from the gravitational potential of dwarf halos; indeed, in dwarfs, the mass, momentum, and metal outflow rates increase with radius whereas energy is roughly conserved, indicating swept up halo gas. Burst-averaged mass loading factors tend to be larger during more powerful star formation episodes and when the inner halo is not virialized, but we see effectively no trend with the dense ISM gas fraction. We discuss how our results can guide future controlled numerical experiments that aim to elucidate the key parameters governing galactic winds and the resulting associated preventative feedback.

astro-ph.GA