RT info:eu-repo/semantics/masterThesis
T1 Deriving Star Formation Histories of Galaxies with Bayesian Deep Learning
A1 Iglesias Navarro, Patricia
A2 Máster Universitario en Astrofísica
AB investigating the properties of these stars, such as their ages, masses, and metallicities, we can gain insightsinto the underlying physical processes that drive the growth and transformation of galaxies over cosmic time,in particular, the triggers and quenching mechanisms of star formation. For this purpose, we explore anamortized implicit inference approach to estimate the posterior distribution of metallicity and non-parametricstar formation histories (SFHs) of galaxies, i.e. star formation rate as a function of cosmic time, using theiroptical spectra. Fed with the spectroscopic predictions of the MILES stellar population models, we generate asample of synthetic SFHs to train and test our model. We show that our approach is capable of reliablyestimating the mass assembly of an integrated stellar population given its optical absorption spectrum with,crucially, well-calibrated uncertainties. Specifically, we achieve 94% accuracy for the time at which a givengalaxy formed 50% of its stellar mass. We apply our pipeline to real observations of very massive ellipticalgalaxies and show that it recovers ranges of SFR(t) consistent with the spectra, as well as the expectedrelation between age and velocity dispersion, demonstrating a good generalization to data. Not only beingable to address a large number of galaxies, but also performing a thick sampling of the posteriors, it allows usto estimate both the deterministic trends and the inherent uncertainty of this highly degenerated inversionproblem, so far inaccessible for more traditional methods. For this reason, we believe that our framework, amachine-learning-based implicit inference applied to full spectral fitting, is remarkably promising to deal withthe size and complexity of upcoming massive spectroscopic surveys such as DESI, WEAVE or 4MOST
YR 2024
FD 2024
LK http://riull.ull.es/xmlui/handle/915/36743
UL http://riull.ull.es/xmlui/handle/915/36743
LA en
DS Repositorio institucional de la Universidad de La Laguna
RD 22-dic-2024