Sugata Kaviraj, Garreth Martin (Arizona), Jim Geach
Galaxies are the building blocks of the observable Universe. Studying their evolution enables us to understand the physical processes that shape our Universe from its earliest epochs to the present day. A fundamental property of galaxies is their ‘morphology’ (i.e. their appearance), which encodes their assembly histories and gives us unique insights into galaxy evolution over cosmic time. For example, mergers between galaxies create ‘tidal debris’ around the remnant, composed of material that ‘splashes out’ during the merger episode. The merger remnants themselves are typically ‘elliptical’ galaxies, which are smooth and featureless, as the gravitational torques during the merger wash out structure in the progenitors. On the other hand, spiral galaxies (such as our own Milky Way) are thought to have had a relatively quiet history, free of galaxy mergers, and shaped largely by the gradual accretion of gas from the cosmic web. Studying the evolution of the morphological mix of galaxies is therefore key to deciphering the physical processes that drive the evolution of the observable Universe over cosmic time.
Astrophysics is entering a ‘Big Data’ era, with new surveys like the Legacy Survey of Space and Time (LSST) and the Hyper Suprime-Cam (HSC) Subaru Strategic Program poised to transform our understanding of galaxy evolution. These surveys will not only image tens of billions of galaxies, but their unparalleled depth will give them unprecedented sensitivity to (the often faint) merger-driven tidal features. However, the sheer size of these datasets makes the classification of galaxy morphology using traditional techniques, e.g. visual inspection of galaxy images by humans, prohibitively time-consuming.
This project will combine (1) state-of-the-art data from HSC and LSST, (2) unsupervised machine-learning techniques for classifying galaxy morphology (e.g. Martin et al. 2020) and (3) in-house cosmological simulations (e.g. Horizon-AGN) to perform the first, automated morphological analysis of the Universe using such surveys. These studies will, in turn, quantify the role of processes such as galaxy mergers and gas accretion in driving star formation, black-hole growth and morphological transformation, in unprecedented detail, over at least 90% of cosmic time.
The student will collaborate closely (through visits and conference trips) with colleagues in Arizona, Paris, Oxford and a worldwide network of scientists within the LSST project (in which our team has a leadership role). While this is an astrophysics project, it is also suitable for students who primarily have a background in computer science which they would like to apply to data intensive astrophysics.