Searches for new high-mass resonances in top-antitop and di-electron final states using the ATLAS detector

The standard model of particle physics (SM) describes all the fundamental particles and their interactions. It is a very successful theory; however, many experimental observations - such as the origin of neutrino mass, particle origin of dark matter, e.t.c. - are either not consistent or not explained by the SM. So, it is inevitable that there has to be a (physics) model beyond the SM which will consistently explain all these observations, not covered by the SM. Several of such extensions predict new heavy particles that can interact with SM particles. This dissertation presents searches for the resonant production of such high-mass particles in dielectron and top-antitop final states. These searches use proton-proton collision data at the center-of-mass energy of 13 TeV collected by the ATLAS detector at the Large Hadron Collider (LHC) between 2015 and 2018. Electrons are stable and easy to reconstruct, but top-quarks decay instantaneously. Two dominant top-decay final states, all-hadronic and semi-leptonic, are studied in this dissertation. The combined mass distributions of all the final-state particles are used to perform model-dependent and model-independent statistical searches. No evidence for the existence of new particles is found in any of the explored final states. Hence, upper limits on production cross-section times branching ratio and lower limits on the mass of heavy Z' particles, predicted by the BSM models, are placed at a 95% confidence level. The dilepton resonance search excludes Z' boson below 3.6 TeV. The resonance search in the boosted all-hadronic top-antitop final state excludes Z' bosons with a mass lower than 4.1 TeV. Whereas in the semi-leptonic search, the same signal is expected to be excluded up to 3.6 TeV. The dissertation also presents a new algorithm for splitting the merged charge clusters in the ATLAS pixel detector, based on a Mixture Density Network (MDN). The performance of this new algorithm is found to be better than the existing algorithm. As a result, the MDN-based algorithm is expected to be used as a default algorithm in ATLAS during the next data collection period, which will start in 2022.Science, Faculty ofPhysics and Astronomy, Department ofGraduat