This thesis presents the results of a search for new physics in proton-proton collisions at the Large Hadron Collider, running with 13 TeV center of mass energy, using data gathered by the Compact Muon Solenoid (CMS). The search targets TeV mass-scale dark matter candidates and uses final states with two opposite-charge and same-flavor light leptons (electrons or muons) having dilepton mass consistent with the Z boson, at least 2 hadronic jets, and at least 100 GeV of transverse momentum imbalance. The thesis contains a historical introduction to particle physics, brief reviews of the standard model and supersymmetry, an in-depth discussion of the acquisition

of data by the Large Hadron Collider and CMS detector, and a pedagogical overview of the analysis methods used. No statistically significant deviation is found from the expected standard model background. The search results are presented and interpreted in the context of several simplified models of supersymmetry, including a model of Gauge Mediated Supersymmetry- Breaking (GSMB) with gluino production and models with Electroweakino production. The excluded mass ranges for these models are advanced by approximately 50-100% with respect to the best previous searches. This work represents the current state of the art for their exclusion

This dissertation presents the first observation of Higgs boson production in association with a top quark-antiquark pair in the diphoton decay channel, with a significance of 6.6 standard deviations. The measurement is performed with a dataset of 13 TeV proton-proton collisions recorded by the Compact Muon Solenoid (CMS) detector at the CERN Large Hadron Collider (LHC), corresponding to an integrated luminosity of 137 \fbinv.

This work presents two searches for new physics characterized by pair-production of strongly interacting particles, each decaying to hadronic jets and a particle that is not detectable. The searches use the full 13 TeV proton-proton collision dataset produced by CERN’s Large Hadron Collider and recorded by the CMS detector from 2016 to 2018, with total integrated luminosity 137 fb−1. The presence of particles interacting too weakly to be detected is inferred using imbalance in the transverse momentum of the collision products, and sensitivity to pair- production is enhanced by requiring large values of the kinematic variable MT2 in events with at least two jets. The first search is inclusive, binning events using the total hadronic transverse energy, the total number of jets, the number of jets reconstructed as originating from a bottom quark, and either the value of MT2 in multijet events, or the transverse momentum of the jet in monojet events. The second search extends the first, by requiring the presence of a disappearing track in the event, and adds binning in the length and transverse momentum of the disappearing track. Both searches are sensitive to a variety of extensions to the Standard Model that include dark matter candidates. Of greatest interest, the results set constraints on pair production of squarks and gluinos as predicted by R-parity conserving supersymmetric extensions of the Standard Model, in which the lightest supersymmetric particle is a neutralino. The first search is sensitive to any decay chain terminating in Standard Model hadrons plus the neutralino, while the second specifically targets, with greatly enhanced sensitivity, decay chains containing an intermediate long-lived chargino. These constraints are the most stringent yet produced by any experiment.

Two related analyses of data from the CMS Experiment are presented. The first is performed using 19.5 inverse femtobarns of proton-proton collision data from CMS Run I. In this analysis, six different asymmetry variables are measured in events with top-antitop quark pairs decaying to final states with two leptons. Unfolding techniques are used to extrapolate these measurements to parton level. No deviations from the predictions of the Standard Model are found, implying the absence of any influence from physics beyond the Standard Model. The second analysis is performed using 35.9 inverse femtobarns of proton-proton collision data collected during CMS Run II. In this analysis, a search is performed for evidence of stop squark pair production and decay to single-lepton final states. Several backgrounds, including dileptonic top-pair production, are estimated using control regions in the data. No excess above the Standard Model backgrounds is found, and exclusion limits are placed on three models of stop squark pair production.

Two projects are presented in this thesis, the first pertaining to a physics analysis using the data from the CMS Experiment at the LHC, and the second pertaining to the development of a new parallelizable algorithm for reconstructing particle tracks in the outer tracker of the CMS Detector at the upcoming High Luminosity LHC.

A search for supersymmetric phenomena beyond the standard model in a final state containing an on-shell Z boson, jets and missing transverse energy is performed using a data sample of proton-proton collisions at $\sqrt{s} = 13$~TeV corresponding to an integrated luminosity of 137 $\ifb$ collected by the CMS Experiment at the LHC between 2016 and 2018. The observed event yields are consistent with standard model predictions in the signal regions. These results are then interpreted to constrain the masses of supersymmetric particles in the context of the signal models. Gluino masses up to 1870 GeV, and chargino (neutralino) masses up to 750 (800) GeV are excluded at the 95\% confidence level, which extends the reach over the previous results by a few hundred GeV.

The High Luminosity LHC (HL-LHC) will increase the instantaneous luminosity by a factor of five larger than the current levels achieved by the LHC. This high pile-up environment requires efficient and fast reconstruction of charged particles. A new algorithm called Line Segment Tracking takes a fundamentally different approach from existing iterative Kalman Filter based algorithms by doing a bottom-up reconstruction of tracks. Track stubs from adjoining detector regions are constructed, and stubs that are consistent with typical track trajectories are hierarchically linked to reconstruct complete tracks. Since the track stubs are produced locally and only require information from neighboring regions, they can be made in parallel. This motivates using architectures like GPUs to take advantage of the parallelism. The algorithm is currently implemented in the context of the CMS Phase-2 Outer Tracker in the HL-LHC, and targets NVIDIA Tesla V100 GPUs. Good physics and timing performance are obtained which paves way for further developments.