Hierarchical Bayesian Modeling of Galaxy SEDs
Abstract
We describe an approach to hierarchically model the SEDs of galaxies with the usual stellar population synthesis models. We frame the problem in the context of the CANDELS dataset, but it can be readily applied to other surveys, datasets, and filters. Unlike the standard maximum-likelihood approach, the hierarchical Bayesian scheme simultaneously models the SEDs of the individual galaxies and the distribution of fitted parameters (e.g. age, extinction, mass). By modeling the system as a whole (ie. individual galaxies and population model), we derive improved estimates for the parameters of the galaxies and population distributions which incorporate parameter correlations, while providing an efficient way of marginalizing over poorly known parameters. Finally, we present a simple simulation as proof-of-concept which highlight the successes of the hierarchical analysis and shortcomings of the traditional approach.
- Publication:
-
American Astronomical Society Meeting Abstracts #221
- Pub Date:
- January 2013
- Bibcode:
- 2013AAS...22114724R