Title: Bayesian matched-filter for detecting sparse stellar populations

Author: Eduardo Balbinot (University of Surrey, UK)

Poster abstract:
We present a method for detecting faint stellar populations in large photometric surveys. The 
technique works by constructing a Poisson mixture model using the colour-magnitude diagram 
(CMD) of a \emph{target} -  ​simple or mixed stellar population​ - and a \emph{background} stellar 
population. We employ a Markov Chain Monte Carlo (MCMC) code to search for the best 
mixture model. An estimate of the number of \emph{target} and \emph{background} objects is 
extracted from the MCMC, as well as their uncertainties. We are also able to extract 
the probability of each individual star belonging to the \emph{target} population, this information 
can be useful for follow-up target selection in future applications. The method easily allows for 
the inclusion of spatial priors (e.g. density profile or stream PDF) as well as the ability to 
include realistic background variations (e.g. spatial variations of the MW background 
stellar population). In this work we show the formalism of the method, simulated tests 
using realistic stellar populations, and a first trial of the method on the well known Palomar 5 
tidal tails using Sloan Digital Survey (SDSS) data. Finally we compare our results to the broadly 
used classic matched-filter technique.