Markov Chain Monte Carlo for Bayesian Parametric Galaxy Modeling in LSST

TL;DR

This paper demonstrates the application of MCMC techniques to Bayesian parametric galaxy modeling for large-scale surveys like LSST, enabling detailed probabilistic inference of galaxy properties from simulated data.

Contribution

It introduces a Bayesian MCMC approach for galaxy modeling in LSST-like data, with a physically informed prior and publicly available implementation.

Findings

Posterior samples enable probabilistic galaxy parameter estimation.

Assessment of the reliability of posterior means and uncertainties.

Implementation of the method in the open-source JIF tool.

Abstract

We apply Markov Chain Monte Carlo (MCMC) to the problem of parametric galaxy modeling, estimating posterior distributions of galaxy properties such as ellipticity and brightness for more than 100,000 images of galaxies taken from DC2, a simulated telescope survey resembling the upcoming Rubin Observatory Legacy Survey of Space and Time (LSST). We use a physically informed prior and apply selection corrections to the likelihood. The resulting posterior samples enable rigorous probabilistic inference of galaxy model parameters and their uncertainties. These posteriors are one key ingredient in a fully probabilistic description of galaxy catalogs, which can ultimately enable a refined Bayesian estimate of cosmological parameters. We systematically examine the reliability of the posterior mean as a point estimator of galaxy parameters, and of the posterior width as a measure of uncertainty, under some common modeling approximations. We implement the probabilistic modeling and MCMC inference using the JIF (Joint Image Framework) tool, which we make freely available online.