Search for vector-like leptons in final states with multiple leptons

The standard model of particle physics stems from the efforts of understanding the ultimate constituents of matter. It has been exceptionally successful in explaining various experimental observations during the last 50 years. Despite being successful, the standard model is incomplete since it fails to answer many open questions. We have failed to develop a successful quantum theory of gravity and connect it to the other three fundamental forces within the framework of the standard model. It is also not understood how the mass of Higgs boson sits at 125 GeV despite it getting significant corrections from all the particles that the Higgs boson couples to. The standard model also does not explain the presence of dark matter in the universe. Many theoretical models have been proposed to account for the inadequacies in the standard model of particle physics. If there is any truth in these propositions, signs of these might show up in the debris of proton-proton collisions in the Large Hadron Collider (LHC) at CERN. The typical way of testing a new theory is to search for one or more hypothesized particles predicted by the theory. Many such efforts have been carried out, but none have found any new particles. While it is possible that the new particles are extraordinarily massive and hence are out of the reach of current energies of the LHC, it is also possible that we are just not looking for the right particles or in the right manner. In this thesis, I have presented a search for an SU(2) doublet vector-like lepton extension of the standard model with couplings to the third generation standard model leptons. The last search for such a doublet was carried out by the L3 collaboration in 2001, where a lower bound of ≈100 GeV was placed on the mass of these particles. The data sample corresponds to 77 fb−1 of integrated luminosity in pp collisions at s=√13 TeV collected by the CMS experiment at the Large Hadron Collider in 2016 and 2017. This is the first search for a vector-like lepton doublet in any of the LHC experiments. Events are primarily categorized based on the multiplicity of light leptons and taus. The missing transverse energy and the scalar sum of transverse momenta of leptons including tau leptons, are used to discriminate the signal model against standard model backgrounds. The observations are consistent with the expectations from the standard model only hypothesis, and the existence of vector-like leptons in the mass range of 150-790 GeV is excluded at 95% confidence level. I have also presented a technique based on convolutional neural networks to improve the sensitivity of the search by better optimizing signal-to-background discrimination. The observations show that this new method provides significant enhancement to the signal-to-background discrimination and also comfortably beats a conventional neural-network based optimization, in most cases.