Computational Approaches to Studying the Stellar Populations of Galaxies

Abstract

Major open questions remain in our understanding of the formation and evolution of galaxies in our universe, and the best tracers of these galactic histories are the stars within. The properties of stellar populations -- the distributions of masses, ages, and metal abundances, both in aggregate and spatially-resolved within galaxies -- are therefore essential to connecting models of galaxy formation to available observations. This thesis discusses two computational methods for connecting the stellar populations of galaxies to their evolutionary histories. First, we investigate the stellar halos of galaxies produced in the Illustris simulation. We present measurements of the gradients in luminosity, metallicity, and age in the inner halos of early-type Illustris galaxies, and discuss their application as tracers of galactic accretion histories. In the simulated galaxies, the steepness of both the luminosity and metallicity profiles are found to be good indicators of the local fraction of ex-situ stars, but that neither is strongly sensitive to the relative contribution from major or minor mergers. Second, we present a new forward-modeling procedure for simultaneously measuring the stellar population properties and distance to semi-resolved galaxies. The resulting product is a GPU-accelerated Python package called PCMDPy, which generates simulated photometry of galaxies in pixel color-magnitude diagram (pCMD) space, and compares the simulations to observed data to infer distances, star formation histories, and metal abundances. We perform extensive mock testing of this code and demonstrate that the new pCMD technique should, in principle, be applicable to galaxies out to nearly 100 Mpc. We then provide the first application of PCMDPy to several nearby (within 20 Mpc) elliptical galaxies. We infer spatially resolved star formation histories within each galaxy, which are suggestive of an inside-out evolution scenario. We also demonstrate the pCMD technique recovers distance estimates accurate to ~20%, with moderate systematic offsets depending on the model complexity. The pCMD technique should prove highly complementary to existing methods of inferring star formation histories, and together the entire history of star formation in the nearby universe can be constrained.