Improving sensitivity with a boosted decision tree in a search for dark photons decaying to lepton jets at CMS

Production of dark photons with GeV-scale mass is predicted in many dark sector models. A distinctive signature is the presence of pairs of oppositely charged leptons (muons only, in this study) where the angle between the leptons in a pair is small so that they form a lepton jet. Discrimination between lepton jets from dark photons and those from background processes is crucial to improving sensitivity. We show the results of a boosted decision tree trained for this purpose and validate its output in control regions.