Abstract

Contributed Talk - Splinter Galaxies

Friday, 11 September 2026, 15:15   (MW-0250)

Building massive galaxies early through efficient star formation and feedback

Akash Vani
MPA

Recent observations suggest that some galaxies assembled their stellar mass rapidly and began quenching at very early times, posing a challenge to simulations calibrated primarily against low-redshift data. In this work, we present a new framework and updated implementation of the semi-analytic model L-Galaxies within the Gadget-4 codebase, incorporating physically motivated and self-consistent prescriptions for key baryonic processes. We introduce a density-modulated star formation model in which the cloud-scale star formation efficiency depends explicitly on the local gas surface density. We also implement an alternative mode of stellar feedback, in which the coupling of stellar winds and supernova feedback to the ISM density and can be strongly suppressed in dense environments. We find that the combination of these updates enhances early stellar mass assembly while preserving agreement with lower-redshift constraints. Together with additional improvements, including updated disc instability prescriptions and a treatment of both wet and dry galaxy mergers, these changes provide a more self-consistent description of star formation, and quenching across cosmic time, in improved agreement with recent observations such as enhanced massive galaxy assembly, improved quenched galaxy mass-size relation and bulge to disk ratios along with boost in quenched galaxies. This talk will present these model developments, discuss the physical mechanisms driving enhanced star formation and early quenching, and introduce a new methodology for calibrating semi-analytic models.