Data‐driven background modelling and trigger algorithms for compressed supersymmetry searches with the CMS experiment at the LHC

Supersymmetry (SUSY) is a highly motivated theory that can provide solutions to central issues and open questions of the standard model (SM) of particle physics. However, the absence of discoveries of exotic particles at the Large Hadron Collider (LHC) puts it under significant pressure. Nonetheless, some corners of the parameter space remain less explored, owing more to their experimental difficulty rather than a lack of theoretical motivation. SUSY scenarios with a compressed mass spectrum, where the mass difference between the produced superpartners and the lightest supersymmetric particle (LSP) is relatively small, offering little detectable energy, are very well motivated by naturalness considerations and dark matter relic constraints. $$$$ A compressed SUSY search focusing on models with top squark (stop) pair-production in the single-lepton channel is presented. Due to the limited available energy and resulting low momentum decay products, such a search is made viable by requiring a boost from an initial-state radiation (ISR) jet. The analysis results are based on collision data from year 2016 of LHC Run 2, recorded with the Compact Muon Solenoid (CMS) detector at ${\sqrt{s} = 13~\mathrm{TeV}}$, corresponding to an integrated luminosity of ${35.9~\mathrm{fb}^{-1}}$. Several data‐driven background modelling methods are implemented for the estimation of sub-leading nonprompt lepton backgrounds, using Monte Carlo (MC) simulation together with normalisation to data. Dedicated studies are conducted for the identification of low momentum electrons as well as improvements in signal modelling. $$$$ Based on the completion of the analysis with 2016 data, another focus of the thesis are improvements of the trigger strategy to extend the acceptance of the signal, also towards less-accessible compressed electroweakino (EWKino) models. Dedicated trigger algorithms are developed, exploiting the typical ISR-boosted signature, in order to achieve lower missing momentum thresholds. The algorithms were used online during year 2018 of LHC Run 2 and are currently collecting data during the ongoing LHC Run 3. A strategy for incorporating the new trigger in the analysis is presented, evaluating the potential expected gains in sensitivity. Possible future improvements in expanding the search capabilities in previously unexplored directions, including recent developments, such as machine learning (ML) or the identification of displaced signatures from long-lived particles (LLP), are also discussed.