Top quark pole mass from $\text{t}\overline{\text{t}}\text{+jet}$ using a machine learning based reconstruction for $\text{t}\overline{\text{t}}$ kinematics

In this thesis, a measurement of the top quark mass as defined in the on-shell renormalization scheme, also known as pole-mass $m_\text{t}^\text{pole}$, is presented. The analysis is performed using events where a top quark-antiquark pair ($\text{t}\overline{\text{t}}$) is produced in association with at least one additional jet ($\text{t}\overline{\text{t}}\text{+jet}$) in proton-proton collisions at the CERN LHC at a center-of-mass energy of 13 TeV. The data are recorded by the Compact Muon Solenoid experiment in 2016, corresponding to a total integrated luminosity of 36.3${\,\text{fb}^\text{-1}}$. Events with two opposite-sign leptons in the final state ($\text{e}^{+}\text{e}^{-}$, $\mathrm{\mu}^{+}\mathrm{\mu}^{-}$, $\text{e}^{\pm}\mathrm{\mu}^{\mp}$) are analyzed. The $\text{t}\overline{\text{t}}\text{+jet}$ production cross section is measured as a function of the inverse of the invariant mass of the $\text{t}\overline{\text{t}}\text{+jet}$ system, $\rho=340\,\text{GeV}/m_{\text{t}\overline{\text{t}}\text{+jet}}$. A novel analysis technique based on machine learning is developed for the reconstruction of the main observable and for the event classification. The measurement is unfolded to the parton level and is compared to the theoretical prediction at next-to-leading order, using the dynamic renormalisation and factorisation scales. The theoretical predictions are obtained by using two alternative sets of parton distribution functions (PDFs). The top quark mass is extracted from the fit of the theoretical predictions to the data. For the ABMP16NLO PDF, this results in $m_\text{t}^\text{pole}=172.93\pm1.36$ GeV. When using the CT18NLO PDF instead, the value is $m_\text{t}^\text{pole} = 172.13\pm1.43$ GeV.