UA Science

You are here

9/20/18: SO/NOAO Joint Colloquium Series: Jennifer Lotz, Space Telescope Science Institute


Title: Quantifying Galaxy Evolution with Simulations and Machine Learning

The assembly of galaxies and their dark matter halos through mergers and accretion is a cornerstone of our current theoretical model of galaxy evolution. With the Sloan Digital Sky Survey and large HST extragalactic surveys, we now have a broad-brush picture of how the structures, star-formation rates, and stellar masses of galaxies have evolved over the past 10 billion years. But we do not understand yet the underlying physical processes connecting observed galaxy structures to their star-formation histories and stellar masses. With cosmological hydrodynamical simulations and numerical techniques for classifying galaxies and tracking their evolution over cosmic time, we seek to understand the physical metamorphosis of massive galaxies in the local Pan-STARRS and distant HST galaxy populations. In particular, we study the disappearance of compact galaxies, the emergence of bulge-dominated galaxies, and the role of galaxy mergers in their evolution. We also analyze simulated images from the ~(100 Mpc)^3 cosmological hydrodynamical simulations Illustris and Illustris TNG. The galaxy merger rates and model distributions of structure, color, and stellar mass are compared to the observed distributions at 0 < z < 3. We find that mock observations from numerical simulations are required to interpret the observed merger and pair fractions, and compute merger timescales. Using extreme deconvolution gaussian mixture models to track the evolution of observed size-mass distributions, we find that galaxies with large sizes and intermediate stellar masses appear most rapidly at z<3, regardless of color. The joint distributions of size/structure, color, and stellar mass place stringent constraints on the numerical simulations, and are not yet fully reproduced.