Étalonnage des jets et mesures précises de sections efficaces de production de jets avec les données de l'expérience ATLAS

The Standard Model (SM) of particle physics describes the elementary particles that constitute matter, as well as three of the four fundamental forces. It remains almost unchallenged by experimental data, thus not indicating any clear breach where New Physics could be located. This thesis focuses on one strategy to test the SM, the comparison of its predictions to precision measurements, and on one aspect of the Quantum Chromo-Dynamics (QCD) section of the SM: jets. The work presented here uses data collected by the ATLAS experiment, located at CERN, from 13 TeV pp collisions produced by the Large Hadron Collider (LHC) during its Run 2 data-taking period. Jets are one of the main observables at the LHC due to their high production cross-section. They are collimated sprays of hadrons, stemming from the hadronization of a quark or gluon. The hadrons produce tracks in the tracking systems of ATLAS and energy deposits in its calorimeters, which then need to be clustered together to form "reconstructed jets". Particle jets and reconstructed jets show differences in their observables, like their transverse momentum pT , due to detector effects which therefore need to be corrected for, to enable comparisons between theory and experiment. Two such strategies are discussed in this thesis: calibration and unfolding, to correct for the detector effects respectively in the scale and the resolution. I have worked on the η-intercalibration, one of the steps of the corrections chain of the Jet Energy Scale (JES). This in situ calibration makes the detector response homogeneous across the whole pseudo-rapidity η range of the detector, by exploiting the transverse momentum pT conservation in dijet systems. I produced the calibration for the 2018 data-taking period for the first time, making all sorts of cross-checks to ensure its quality. I also worked on the "absolute" implementation of this method, that aims at disentangling the physics effects from the detector effects, to better understand them separately. This led to the reduction by 50 % of the modeling uncertainty that was dominant in the 25 to 40 GeV bin. It also allowed me to make the important verification at particle level that the uncertainties designed to cover for physics effects are not underestimated. Furthermore, I contributed to the inclusive and leading jet production cross-section measurements, two doubledifferential precision measurements that count jets in bins of pT and rapidity |y|. While the inclusive measurement counts every single jet produced above a certain pT threshold, the leading jet measurement only counts one per event: the one with the highest pT . The Jet Energy Resolution (JER) effects are corrected for by the Iterative Dynamically Stabilized unfolding (IDS) method. Systematic uncertainties on the JES and JER have been propagated through the unfolding, using the Bootstrap method to evaluate the statistical uncertainties as well as correlations. I have improved the evaluation of the bias related to the sensitivity of the unfolding method to the shape differences between data and MC, and I evaluated the impact of the non-Gaussianity of the JER in a new uncertainty. Results of the inclusive jet production cross-section are shown and compared to theoretical predictions in this thesis, exhibiting an important disagreement that is being addressed by multiple strategies, like the use of non-smoothed statistical uncertainties of the η-intercalibration