Searching for New Physics in Dilepton Events - A Study of the Photon-Induced Process

This Master thesis presents an analysis of the data collected during the full Run II (2016, 2017 and 2018) by the CMS detector, located at the proton-proton collider, the LHC, at CERN. Events with high energy dileptons are studied, with a focus on photon-induced (PI) lepton pair production. Protons, at high energy, are formed by valence quarks ($u$ and $d$) and an intricate sea of quarks, gluons, and photons, described by the parton distributions functions. In proton-proton collisions at the LHC, two of these photons can interact to form a pair of leptons: the PI process. However, other processes such as the Drell-Yan (DY), the $t\bar{t}$ and the diboson $WW$ also contribute to high energy dilepton final states and dominate the PI. Exploiting differences at the event level, the PI is selected from these backgrounds by demanding a low missing transverse energy significance, no $b$-jets, at most a single jet, and a low number of charged hadrons in the vertex fit. These conditions emanate from the absence of neutrinos and quarks in the final state as well as the relative stability of the protons in a PI interaction. The selected signal estimated using PI simulation is fitted to the data to deliver mass-dependent corrections to the theoretical cross section. Two procedures are implemented. The first one uses the difference in topology between the PI and the DY in the $\cos \theta$ distributions, due to the forward-backward asymmetry of the latter. The second one relies on the number of charged hadrons from the vertex fit. The corrections are found to be in good agreement with recently published results, pointing out an overestimation originating from finite size effects in the proton being neglected. A BSM search is then introduced by displaying the invariant mass spectrum in the PI enriched region. No significant deviations are observed within statistical incertitude, though a deeper study is needed to properly account for systematics.