| Abstract |
The Nancy Grace Roman Space Telescope’s deep, high-resolution imaging and wide field-of-view enable unprecedented statistical analyses of galaxy mergers as drivers of “the growth of cosmic structure,” which is a stated Roman science objective. Using machine learning trained on galaxies in three different cosmological simulations, we will identify galaxy mergers in the 100 million galaxies at 0.3 ≤ z ≤ 3 in the High-Latitude Wide-Area Survey Medium Tier Field 1. For added scientific value, our Roman Galaxy Morphology and Merger Catalog will be the first machine learning-based merger catalog to include classifications of the mergers as both 1) pre- or post-coalescence, and 2) minor or major merger. We will use our catalog to measure the evolution of the galaxy merger fraction with redshift, which has been previously poorly constrained. With this measurement, we will uncover the relative roles of minor and major mergers (as opposed to other mechanisms such as star formation) in the mass assembly of galaxies over cosmic time.
Through analysis of Medium Tier and Deep Tier data, we will also carry out the first large scale study of the best practices for imaging-based merger identification in general. We will determine whether restframe ultraviolet imaging helps or hinders merger identification, and what biases are introduced to merger identification when nonuniform restframe wavelength imaging is used. The results of these analyses will steer future merger classifications in any galaxy imaging survey carried out with any telescope, and we will place particular emphasis on laying a roadmap for future merger identification in the Roman Wide Tier (including the use of Rubin/LSST bands to supplement the one band of Roman Wide Tier imaging) when that Tier is observed. |