Analysis of stellar evolution models for galaxy formation simulations

Our theoretical understanding of galaxy formation relies heavily on numerical simulations to model the evolution of these systems over cosmic time. Most state-of-the-art simulations incorporate stellar evolution to allow galaxies to evolve self-consistently. However, since resolving single stars in galaxy-scale simulations is computationally unfeasible, stellar evolution must be represented through effective prescriptions at scales below the resolution of the simulation, an approach known as “sub-grid physics”. In this thesis, we investigate how stellar evolution is implemented in two widely used galaxy formation models: SMUGGLE (Marinacci et al., 2019) and FIRE-3 (Hopkins et al., 2022). We compare and characterize the key features of their respective stellar evolution prescriptions and explore the potential implications that these differences have on the evolution of the interstellar medium (ISM) in terms of mass, chemical composition and star formation. We then use the stellar population code SEVN (Iorio et al., 2023) to develop a new stellar evolution module for the SMUGGLE model. To achieve this, we create a software pipeline that replicates SMUGGLE implementation of stellar evolution and, using input obtained from SEVN, generates updated stellar evolution models. This pipeline can be applied to other simulation frameworks that share the same stellar evolution infrastructure as SMUGGLE, such as the Illustris and IllustrisTNG models. Finally, we compare the original SMUGGLE stellar evolution model with the updated version derived from SEVN, finding that the newer model predicts a lower mass return to the ISM, but with a higher level of metal enrichment. We highlight that pipeline developed in this thesis has the potential to facilitate the generation of future stellar evolution models and to assist in analyzing the results of galaxy simulations, especially when a precise determination of the stellar mass return and total metal production are needed.