Validation of control regions for a data-based background estimate for a dark photon search at CMS

Dark photons which are produced from the decay of massive particles will be highly collimated, forming distinctive lepton jets which have a low background. However, identification of control regions that are dominated by background is crucial to estimating the background in the signal regions. We demonstrate several possible control regions, based on invariant mass, primary vertex, and neural network output, and evaluate their usefulness in estimating the background yield in the signal region.