Pop-cosmos: Insights from Generative Modeling of a Deep, Infrared-selected Galaxy Population

Image simulations are essential tools for preparing and validating the analysis of current and future wide-field optical surveys. However, the galaxy models used as the basis for these simulations are typically limited to simple parametric light profiles, or use a fairly limited amount of available space-based data. In this work, we propose a methodology based on deep generative models to create complex models of galaxy morphologies that may meet the image simulation needs of upcoming surveys. We address the technical challenges associated with learning this morphology model from noisy and point spread function (PSF)-convolved images by building a hybrid Deep Learning/physical Bayesian hierarchical model for observed images, explicitly accounting for the PSF and noise properties. The generative model is further made conditional on physical galaxy parameters, to allow for sampling new light profiles from specific galaxy populations. We demonstrate our ability to train and sample from such a model on galaxy postage stamps from the HST/ACS COSMOS survey, and validate the quality of the model using a range of second- and higher order morphology statistics. Using this set of statistics, we demonstrate significantly more realistic morphologies using these deep generative models compared to conventional parametric models. To help make these generative models practical tools for the community, we introduce galsim-hub, a community-driven repository of generative models, and a framework for incorporating generative models within the galsim image simulation software.

We present results for the assembly and star formation histories of massive (~L*) red sequence galaxies in 11 spectroscopically confirmed, infrared-selected galaxy clusters at 1.0 < z < 1.5, the precursors to present-day massive clusters with M ~ 10^15 M_sun. Using rest-frame optical photometry, we investigate evolution in the color and scatter of the red sequence galaxy population, comparing with models of possible star formation histories. In contrast to studies of central cluster galaxies at lower redshift (z < 1), these data are clearly inconsistent with the continued evolution of stars formed and assembled primarily at a single, much-earlier time. Specifically, we find that the colors of massive cluster galaxies at z = 1.5 imply that the bulk of star formation occurred at z ~ 3, whereas by z = 1 their colors imply formation at z ~ 2; therefore these galaxies exhibit approximately the same luminosity-weighted stellar age at 1 < z < 1.5. This likely reflects star formation that occurs over an extended period, the effects of significant progenitor bias, or both. Our results generally indicate that massive cluster galaxy populations began forming a significant mass of stars at z >~ 4, contained some red spheroids by z ~ 1.5, and were actively assembling much of their final mass during 1 < z < 2 in the form of younger stars. Qualitatively, the slopes of the cluster color-magnitude relations are consistent with no significant evolution relative to local clusters.

view Abstract Citations (153) References (194) Co-Reads Similar Papers Volume Content Graphics Metrics Export Citation NASA/ADS Radio Identification of UGC Galaxies: Starbursts and Monsters Condon, J. J. ; Broderick, J. J. Abstract Radio identifications of galaxies in the Uppsala General Catalogue of Galaxies with δ < +82^deg^ were made from the Green Bank 1400 MHz sky maps. Every source having peak flux density S_P_ >= 150 mJy in the ~12 arcmin FWHM map point-source response and position < 5 arcmin in both coordinates from the optical position of any UGC galaxy was considered a candidate identification to ensure that very extended (up to 1 Mpc) and asymmetric sources would not be missed. Maps in the literature or new 1.49 GHz VLA C array maps made with 18 arcsec resolution were used to confirm or reject candidate identifications. The resulting list of 176 confirmed identifications should be complete, reliable, and suitable for statistical investigations of radio emission from nearby (D < 300 Mpc for H_0_ = 50 km s^-1^ Mpc^-1^) galaxies of all morphological types. The distribution of the infrared-radio parameter u=log (S_60 micron_/S_1400 MHz_) for most radio-selected spirals is roughly Gaussian with mean <u> = +2.02 +/- 0.02 and rms width σ_u_ = 0.20. These values agree with <u> = +2.15 and σ_u_ <= 0.3 found for infrared- selected galaxies since a difference {DELTA}u~(3/2) ln (10)σ^2^_u_ = 0.14 is expected from frequency selection alone. The small observed value of {DELTA}u implies that populations of radio-selected spiral galaxies are very similar to infrared-selected galaxies, not "normal" (optically selected) galaxies. Radio sources powered by "starbursts" can be effectively distinguished from those whose energy sources are "monsters" (e.g., super- massive black holes) by three radio and infrared criteria: ( 1) radio morphology, (2) u>= 1.6, and (3) infrared spectral index α_IR_=log (S_60 micron_/S_25 micron_)/log (60/25) >= +1.25. Publication: The Astronomical Journal Pub Date: July 1988 DOI: 10.1086/114788 Bibcode: 1988AJ.....96...30C Keywords: Astronomical Catalogs; Black Holes (Astronomy); Radio Galaxies; Spiral Galaxies; Starburst Galaxies; Very Large Array (Vla); Galactic Clusters; Sky Surveys (Astronomy); Astrophysics; RADIO SOURCES: GALAXIES; GALAXIES: SEYFERTS full text sources ADS | data products NED (186) SIMBAD (179)

In order to prepare for the upcoming wide-field cosmological surveys, large simulations of the Universe with realistic galaxy populations are required. In particular, the tendency of galaxies to naturally align towards overdensities, an effect called intrinsic alignments (IA), can be a major source of systematics in the weak lensing analysis. As the details of galaxy formation and evolution relevant to IA cannot be simulated in practice on such volumes, we propose as an alternative a Deep Generative Model. This model is trained on the IllustrisTNG-100 simulation and is capable of sampling the orientations of a population of galaxies so as to recover the correct alignments. In our approach, we model the cosmic web as a set of graphs, where the graphs are constructed for each halo, and galaxy orientations as a signal on those graphs. The generative model is implemented on a Generative Adversarial Network architecture and uses specifically designed Graph-Convolutional Networks sensitive to the relative 3D positions of the vertices. Given (sub)halo masses and tidal fields, the model is able to learn and predict scalar features such as galaxy and dark matter subhalo shapes; and more importantly, vector features such as the 3D orientation of the major axis of the ellipsoid and the complex 2D ellipticities. For correlations of 3D orientations the model is in good quantitative agreement with the measured values from the simulation, except for at very small and transition scales. For correlations of 2D ellipticities, the model is in good quantitative agreement with the measured values from the simulation on all scales. Additionally, the model is able to capture the dependence of IA on mass, morphological type, and central/satellite type.

Forthcoming cosmological imaging surveys, such as the Rubin Observatory LSST, require large-scale simulations encompassing realistic galaxy populations for a variety of scientific applications. Of particular concern is the phenomenon of intrinsic alignments (IA), whereby galaxies orient themselves towards overdensities, potentially introducing significant systematic biases in weak gravitational lensing analyses if they are not properly modeled. Due to computational constraints, simulating the intricate details of galaxy formation and evolution relevant to IA across vast volumes is impractical. As an alternative, we propose a Deep Generative Model trained on the IllustrisTNG-100 simulation to sample 3D galaxy shapes and orientations along with correlated scalar features, conditioned on the tidal fields and halo mass. The architecture consists of a SO(3) $\times \mathbb {R}^n$ diffusion generative model, implemented with E(3) equivariant Graph Neural Networks that explicitly respect the Euclidean symmetries of our Universe. The generated and the true values for geometric quantities using two-point statistics are statistically consistent; e.g., Wasserstein-1 distances indicate percent-level (or better) agreement in the 1D distributions of scalar quantities. Notably, our model demonstrates the ability to jointly model Euclidean-valued scalars (galaxy sizes, shapes, and colors) along with non-Euclidean valued SO(3) quantities (galaxy orientations) that are governed by highly complex galactic physics at non-linear scales.