UniverseMachine: Predicting Galaxy Star Formation over Seven Decades of Halo Mass with Zoom-in Simulations

We apply the empirical galaxy--halo connection model UniverseMachine to dark matter-only zoom-in simulations of isolated Milky Way (MW)--mass halos along with their parent cosmological simulations. This application extends textsc{UniverseMachine} predictions into the ultra-faint dwarf galaxy regime ($ 10^{2},mathrm{M_{odot}} leqslant M_{ast} leqslant 10^{5},mathrm{M_{odot}}$) and yields a well-resolved stellar mass--halo mass (SMHM) relation over the peak halo mass range $10^8,mathrm{M_{odot}}$ to $10^{15},mathrm{M_{odot}}$. The extensive dynamic range provided by the zoom-in simulations allows us to assess specific aspects of dwarf galaxy evolution predicted by textsc{UniverseMachine}. In particular, although UniverseMachine is not constrained for dwarf galaxies with $M_* lesssim 10^{8},mathrm{M_{odot}}$, our predicted SMHM relation is consistent with that inferred for MW satellite galaxies at $z=0$ using abundance matching. However, UniverseMachine predicts that nearly all galaxies are actively star forming below $M_{ast}sim 10^{7},mathrm{M_{odot}}$ and that these systems typically form more than half of their stars at $zlesssim 4$, which is discrepant with the star formation histories of Local Group dwarf galaxies that favor early quenching. This indicates that the current UniverseMachine model does not fully capture galaxy quenching physics at the low-mass end. We highlight specific improvements necessary to incorporate environmental and reionization-driven quenching for dwarf galaxies, and provide a new tool to connect dark matter accretion to star formation over the full dynamic range that hosts galaxies.

Dark Matter Constraints from a Unified Analysis of Strong Gravitational Lenses and Milky Way Satellite Galaxies

Joint analyses of small-scale cosmological structure probes are relatively unexplored and promise to advance measurements of microphysical dark matter properties using heterogeneous data. Here, we present a multidimensional analysis of dark matter substructure using strong gravitational lenses and the Milky Way (MW) satellite galaxy population, accounting for degeneracies in model predictions and using covariances in the constraining power of these individual probes for the first time. We simultaneously infer the projected subhalo number density and the half-mode mass describing the suppression of the subhalo mass function in thermal relic warm dark matter (WDM), $M_{mathrm{hm}}$, using the semianalytic model $mathrm{texttt{Galacticus}}$ to connect the subhalo population inferred from MW satellite observations to the strong lensing host halo mass and redshift regime. Combining MW satellite and strong lensing posteriors in this parameter space yields $M_{mathrm{hm}}<10^{7.0} M_{mathrm{odot}}$ (WDM particle mass $m_{mathrm{WDM}}>9.7 mathrm{keV}$) at $95%$ confidence and disfavors $M_{mathrm{hm}}=10^{7.4} M_{mathrm{odot}}$ ($m_{mathrm{WDM}}=7.4 mathrm{keV}$) with a 20:1 marginal likelihood ratio, improving limits on $m_{mathrm{WDM}}$ set by the two methods independently by $sim 30%$. These results are marginalized over the line-of-sight contribution to the strong lensing signal, the mass of the MW host halo, and the efficiency of subhalo disruption due to baryons and are robust to differences in the disruption efficiency between the MW and strong lensing regimes at the $sim 10%$ level. This work paves the way for unified analyses of next-generation small-scale structure measurements covering a wide range of scales and redshifts.

Searching for Dwarf Galaxies in ${it Gaia}$ DR2 Phase-Space Data Using Wavelet Transforms

We present a wavelet-based algorithm to identify dwarf galaxies in the Milky Way in ${it Gaia}$ DR2 data. Our algorithm detects overdensities in 4D position--proper motion space, making it the first search to explicitly use velocity information to search for dwarf galaxy candidates. We optimize our algorithm and quantify its performance by searching for mock dwarfs injected into ${it Gaia}$ DR2 data and for known Milky Way satellite galaxies. Comparing our results with previous photometric searches, we find that our search is sensitive to undiscovered systems at Galactic latitudes~$lvert brvert>20^{circ}$ and with half-light radii larger than the 50% detection efficiency threshold for Pan-STARRS1 (PS1) at (${it i}$) absolute magnitudes of =$-7<M_V<-3$ and distances of $32$ kpc $< D < 64$ kpc, and (${it ii}$) $M_V< -4$ and $64$ kpc $< D < 128$ kpc. Based on these results, we predict that our search is expected to discover $5 pm 2$ new satellite galaxies: four in the PS1 footprint and one outside the Dark Energy Survey and PS1 footprints. We apply our algorithm to the ${it Gaia}$ DR2 dataset and recover $sim 830$ high-significance candidates, out of which we identify a gold standard list of $sim 200$ candidates based on cross-matching with potential candidates identified in a preliminary search using ${it Gaia}$ EDR3 data. All of our candidate lists are publicly distributed for future follow-up studies. We show that improvements in astrometric measurements provided by ${it Gaia}$ EDR3 increase the sensitivity of this technique; we plan to continue to refine our candidate list using future data releases.

comp-syn: Perceptually Grounded Word Embeddings with Color

Popular approaches to natural language processing create word embeddings based on textual co-occurrence patterns, but often ignore embodied, sensory aspects of language. Here, we introduce the Python package comp-syn, which provides grounded word embeddings based on the perceptually uniform color distributions of Google Image search results. We demonstrate that comp-syn significantly enriches models of distributional semantics. In particular, we show that (1) comp-syn predicts human judgments of word concreteness with greater accuracy and in a more interpretable fashion than word2vec using low-dimensional word-color embeddings, and (2) comp-syn performs comparably to word2vec on a metaphorical vs. literal word-pair classification task. comp-syn is open-source on PyPi and is compatible with mainstream machine-learning Python packages. Our package release includes word-color embeddings for over 40,000 English words, each associated with crowd-sourced word concreteness judgments.

Constraints on the epoch of dark matter formation from Milky Way satellites

A small fraction of thermalized dark radiation that transitions into cold dark matter (CDM) between big bang nucleosynthesis and matter-radiation equality can account for the entire dark matter relic density. Because of its transition from dark radiation, late-forming dark matter (LFDM) suppresses the growth of linear matter perturbations and imprints the oscillatory signatures of dark radiation perturbations on small scales. The cutoff scale in the linear matter power spectrum is set by the redshift $z_T$ of the phase transition; tracers of small-scale structure can therefore be used to infer the LFDM formation epoch. Here, we use a forward model of the Milky Way (MW) satellite galaxy population to address the question: How late can dark matter form? For dark radiation with strong self-interactions, which arises in theories of neutrinolike LFDM, we report $z_{T}>5.5times 10^6$ at $95%$ confidence based on the abundance of known MW satellite galaxies. This limit rigorously accounts for observational incompleteness corrections, marginalizes over uncertainties in the connection between dwarf galaxies and dark matter halos, and improves upon galaxy clustering and Lyman-$alpha$ forest constraints by nearly an order of magnitude. We show that this limit can also be interpreted as a lower bound on $z_T$ for LFDM that free-streams prior to its phase transition, although dedicated simulations will be needed to analyze this case in detail. Thus, dark matter created by a transition from dark radiation must form no later than one week after the big bang.