Detecting Galaxy Clusters in the DLS and CARS: a Bayesian Cluster Finder

Abstract: The detection of galaxy clusters in present and future surveys enables measuring mass-to-light ratios, clustering properties or galaxy cluster abundances and therefore, constraining cosmological parameters. We present a new technique for detecting galaxy clusters, which is based on the Matched Filter Algorithm from a Bayesian point of view. The method is able to determine the position, redshift and richness of the cluster through the maximization of a filter depending on galaxy luminosity, density and photometric redshift combined with a galaxy cluster prior. We tested the algorithm through realistic mock galaxy catalogs, revealing that the detections are 100% complete and 80% pure for clusters up to z <1.2 and richer than \Lambda > 25 (Abell Richness > 0). We applied the algorithm to the CFHTLS Archive Research Survey (CARS) data, recovering similar detections as previously published using the same data plus additional clusters that are very probably real. We also applied this algorithm to the Deep Lens Survey (DLS), obtaining the first sample of optical-selected galaxy in this survey. The sample is complete up to redshift 0.7 and we detect more than 780 cluster candidates up to redshift 1.2. We conclude by discussing the differences between previous weak lensing detections in this survey and optical detections in both samples.