Étude du couplage du boson de Higgs au quark top dans les canaux avec deux leptons de même signe avec l'expérience ATLAS au LHC

The Large Hadron Collider (LHC) at CERN (European Organization for Nuclear Research) enables physics exploration at subatomic level. After shutting down for planned maintenance in early 2013, the LHC restarted in June 2015, increasing luminosity and centre of mass energy from 8 to 13 TeV. The first part of this document is dedicated to the electron reconstruction efficiency measurement with the ATLAS detector using $27.7\,\text{fb}^{-1}$ of data collected in 2016. In the intense hadronic environment, leptons are markers of interest for many processes. Precise measurements of their reconstruction efficiency are then crucial. Measurement method and background estimates are detailed for electrons, and results are computed in both data and Monte Carlo simulation samples. Electron reconstruction efficiency varies from 97\,\% to 99\,\%. The measurement accuracy is at the percent and per mille level at low energy and $E_T>20\,\text{GeV}$, respectively. Those results have been used in all ATLAS physics analyses involving electrons. The second part of this manuscript focuses on the Higgs boson coupling $y_t$ to the top quark. The analysis aims at a direct measurement of the coupling $y_t$ from the associated production of the Higgs boson with a top quark-antiquark pair ($t\bar{t}H$). Seven multileptonic channels, mainly targetting Higgs boson decays $H\rightarrow WW^*$ and $H\rightarrow\tau\tau$, have been studied using $36.1\,\text{fb}^{-1}$ of proton-proton collision data collected in 2015 and 2016. This document focuses on the most sensitive one for the measurement, characterized by a final state with two same-charge leptons. Background discrimination is based on machine learning with multivariate analysis (MVA) techniques. The optimisation strategy is described and results detailed for the $2\ell\mathrm{SS}$ channel. Combining all $t\bar{t}H$ studied decay channels, the ATLAS and CMS experiments independently claimed an evidence, and then observation for the $t\bar{t}H$ process. The same methodology has been used in the search for flavour changing neutral currents in top-quark decays, forbidden at tree level. The $2\ell\mathrm{SS}$ and $3\ell$ channels are optimized and their results are combined. The current best upper limits on the branching ratios $\mathcal{B}(t\rightarrow Hu)$ and $\mathcal{B}(t\rightarrow Hc)$ have been set to 0.16\,\% and 0.19\,\%, respectively.