Search for Higgs boson production in association with a top quark pair in two same-sign lepton final states with the ATLAS detector at LHC

The Large Hadron Collider (LHC) at CERN restarted in spring 2015 for three years (Run2) at an unexplored center-of-mass energy of 13 TeV; An ideal place to search for physics beyond the Standard Model such as supersymmetry (SUSY) and extra dimensions. A precise measurement of electron reconstruction efficiency in ATLAS, one of the two general purpose experiments of the LHC, is presented in the first part of this thesis with $Z\rightarrow ee$ data sample using 3.2 $fb^{-1}$ of data recorded in 2015. This allows to extract scale factors between data and simulation that are used by all ATLAS physics analyses involving electrons. The results show the high ability of the ATLAS detector to reconstruct electrons from one hand and the good understanding of its performance on the other hand. The second part of the thesis is dedicated to a search for the Higgs boson production in association with a top quark pair ($t\bar{t}H$), which could allow a first direct measurement of the top quark Yukawa coupling and could reveal new physics. The signature with two same-charge light leptons (electron or muon) without a hadronically decaying tau lepton final state, targeting the decays $H\rightarrow$ $WW^{\star}$, is examined using the first 10% of the total expected Run2 dataset. Events with fake (non-prompt) leptons represent the main reducible background of this signature. The estimation of this background largely drives the signal sensitivity. An improved method to estimate it has been developped and is discussed in details in this thesis. Driving the total error, fake leptons background is found to be 1.5 to 3.6 times higher than in simulation and represent between 32 and 48% of the total background. The best-fit value of the ratio of observed and Standard Model cross sections of $t\bar{t}H$ production process, combining with other multilepton channels, is 2.5$\pm$ 0.7 (stat) $^{+1.1}_{-0.9}$ (syst), and an upper limit on this ratio of 4.9 (2.3 expected) is found at 95% confidence level.