Measurement of $W$ boson production in $pp$ collisions at 5.02 TeV and optimisation of electron identification in Pb+Pb collisions with the ATLAS detector

Precise measurements of electroweak boson production at hadron colliders provide stringent tests of Standard Model predictions, including both the electroweak theory and quantum chromodynamics. Due to the strong correlation between the weak boson rapidity and the momentum fraction of initial-state partons, these measurements can be used to constrain parton distribution functions of the proton. The work presented in this thesis is focused on the measurement of $W$ boson production in proton-proton collisions at $\sqrt{s} = 5.02$ TeV using data collected by the ATLAS experiment. It is the first measurement of $W$ boson production cross-sections at this collision energy. The thesis also presents an optimisation of the ATLAS electron identification developed for data taken in lead-lead collisions. Both parts of the thesis provide important inputs to measurements of $W$ boson production in heavy-ion collisions at $\sqrt{s_\text{NN}} = 5.02$ TeV. Cross-sections for $W^\pm\to\ell^\pm\nu$ production are measured in the muon and electron decay channels. In order to achieve a satisfying precision of the cross-section measurements, it is necessary to study in detail the efficiencies of lepton reconstruction, identification, trigger and isolation criteria, as well as the lepton energy and momentum calibrations. Detailed studies related to the muon performance are presented in the thesis. Another important part of the measurement is related to the evaluation of background processes which contribute to the observed distributions. Contributions from electroweak boson and top-quark production are evaluated using Monte Carlo simulations, while the QCD multi-jet background yield is estimated with a data-driven technique. After background subtraction and correction for detector effects, the resulting cross-sections measured separately in the muon and electron decay channels are found to agree within uncertainties, and are therefore combined accounting for uncertainty correlations. The reported $W^\pm\to\ell^\pm\nu$ production cross-sections are measured in a fiducial phase-space region restricted by requirements on the lepton transverse momentum, $p_\text{t}^\ell > 25$ GeV, absolute lepton pseudorapidity, $|\eta_\ell|<2.5$, neutrino transverse momentum, $p_\text{t}^\nu > 25$ GeV, and transverse mass, $m_\text{t} > 40$ GeV. Cross-sections integrated over this region are measured to be: \begin{equation} \sigma_{W^+\to\ell^+\nu} = 2266 \pm 9~(\text{stat}) \pm 29~(\text{syst}) \pm 43~(\text{lumi})~\text{pb} \nonumber \end{equation} \begin{equation} \sigma_{W^-\to\ell^-\nu} = 1401 \pm 7~(\text{stat}) \pm 18~(\text{syst}) \pm 27~(\text{lumi})~\text{pb} \nonumber \end{equation} with a precision of 1.3-1.4%, excluding the luminosity uncertainty. Differential $W^\pm\to\ell^\pm\nu$ production cross-sections are measured as a function of $|\eta_\ell|$ with measurement uncertainties ranging between 2.2% and 3.6%, while the lepton charge asymmetry is measured with an absolute uncertainty at the level of $1 \cdot 10^{-3}$. In general, a satisfying measurement precision is reached, even though the size of the dataset is much smaller compared to the samples used for high-precision LHC measurements of electroweak boson production at higher energies. Heavy-ion collisions at LHC energies are characterised by a large hadronic activity. The most central collisions produce thousands of particles from the underlying event, which lead to increased detector occupancies with respect to those observed in proton-proton collisions. The large occupancies result in a performance degradation of particle reconstruction and identification algorithms. In particular, the electron identification algorithms used by ATLAS in proton-proton collisions suffer from efficiency losses due to distorted distributions of calorimetric variables. Therefore, a dedicated optimisation of electron identification criteria for lead--lead collisions is necessary. The optimisation is performed for the likelihood-based identification algorithm. It is based on a high-level discriminant computed from variables related to the calorimeter cluster and the charged-particle track that constitute the reconstructed electron, and to the cluster-track matching. The distributions of electron identification variables are determined as a function of the lead-lead collision centrality in intervals of $|\eta_e|$ and $p_\text{t}^e$. Then, the likelihood discriminant is re-evaluated based on these inputs, and new centrality-dependent identification criteria are developed. Following the optimisation, two sets of identification criteria are validated using data to verify that they indeed recover efficiency losses for the most central events. After successful validation, they are integrated with the ATLAS reconstruction software. The high and almost centrality-independent efficiency of the optimised identification criteria is confirmed in measurements of $W$ and $Z$ boson production in lead-lead collisions at $\sqrt{s_\text{NN}} = 5.02$ TeV.