Optimizing the photon selection of the CMS Single-Photon search for Supersymmetry using multivariate analyses

The purpose of this thesis is to improve the photon selection of the CMS SinglePhoton search for Supersymmetry by using multivariate analyses.The Single-Photon search aims to find Supersymmetry (SUSY) in data taken by theCompact Muon Solenoid (CMS) detector at the Large Hadron Collider located atthe research center CERN. SUSY is an extension of the standard model of particlephysics. The search is designed for a general gauge mediation scenario, which describes the gauge mediated SUSY breaking. The analysis uses final states with jets,at least one photon and missing transverse energy. A data-driven prediction of themultijet background is performed for the analysis. For this purpose, photon candidates have to be classified into two selections.In this thesis the usage of multivariate analyses for the photon candidate classification is studied. The methods used are Fisher Discriminant, Boosted Decision Treesand Artificial Neural Networks. Their performance is evaluated with respect to different aspects important for the Single-Photon search and they are compared to theanalysis’ current selection method. It is found, that the studied approaches offersome advantages like an increased photon identification efficiency, a reduced statistical background uncertainty and a larger flexibility in the choice of the backgroundcontrol region.