Search for dark matter with machine learning techniques in leptonic final states and missing transverse energy at the CMS experiment

This thesis presents work for the Compact Muon Solenoid (CMS) experiment at the Large Hadron Collider (LHC). The two main subjects are a search for Dark Matter events in final states with one lepton within two supersymmetry (SUSY) models and studies on the performance of the Hadron Outer (HO) calorimeter as a support for the muon system and the muon trigger. In this thesis, results are presented from a search for supersymmetry in events with a single electron or muon and hadronic jets. The data correspond to a sample of proton-proton collisions at $\sqrt{s} = 13$ TeV with an integrated luminosity of 137 $fb^{-1}$, recorded in 2016, 2017, and 2018 by the CMS experiment. SUSY is one of the most predictive theories for new physics and contains a broad variety of potential new particles. Exclusive search regions are defined with the help of a multi-class Deep Neural Network (DNN) classifier specifically developed for this analysis. This multi-class network is used as a data-driven background estimator. The main Standard Model backgrounds are fixed in this approach using several rate parameters which are measured in exclusive control regions. The numbers of observed events are consistent with the expectations from Standard Model processes and the results are used to set lower limits on supersymmetric particle masses in the context of two simplified models of gluino pair production. In the first model, where each gluino decays to a top quark-antiquark pair and a neutralino, gluino masses up to 2.25 TeV are excluded at 95\% Confidence Level (CL). The second model considers a three-body decay to a light quark-antiquark pair and a chargino, which subsequently decays to a W boson and a neutralino. In this model, gluinos are excluded up to 2.3 TeV. Detector studies were performed to demonstrate and to quantify the ability of the HO to support the muon system. These studies are based on data collected by the CMS detector in 2017. The HO can support the muon system in many different aspects, especially by providing information for the Level-1 trigger. The ability of the HO to replace the first muon station in case of failure was studied. Furthermore, it is shown that the HO can cover the wheel gaps that are known to be a source of inefficiency in the muon detector. An additional study demonstrates that the HO can also provide vital information to reject misidentified muons with high transverse momentum. In addition to these studies, this thesis documents the performed upgrade work for the HO backend electronics. Old VME-based backend electronics were replaced with modern $\mu$TCA-based electronics