The shape of the Higgs: The inclusive and differential measurement of the fiducial Higgs boson production cross-section in the diphoton decay channel at the CMS detector and the reliability of legacy components used for the HL-LHC upgrade of the CMS electromagnetic calorimeter

This thesis presents the inclusive and differential measurement of the fiducial Higgs boson production cross-section in the diphoton decay channel with the CMS detector. A novel machine learning-based technique is introduced to effectively correct the detector simulation of electromagnetic clusters in the CMS electromagnetic calorimeter. Finally, an analysis to ensure the reliability of legacy components of the barrel part of the electromagnetic calorimeter until the end of the High Luminosity LHC is explained. The Higgs to diphoton cross-section is measured inclusively in a phase space region closely following the acceptance of the CMS detector and differentially depending on 20 observables covering the kinematics of the diphoton pair and additional jets, angular variables of and between the dijet and diphoton systems and event-level observables. Additionally, the $\text{H}\rightarrow\gamma\gamma$ cross-section is measured double differentially with respect to two sets of two observables. The set of measurements is complemented by measuring the cross-section in four regions of the phase space targeting different Higgs boson production modes. The inclusive fiducial \Hgg cross-section is measured to $\sigma_{\mathrm{fid}}=73.4_{-5.9}^{+6.1}~\text{fb}$ and is, as all other results of the analysis, compatible with the Standard Model prediction. The effective simulation correction method uses a newly developed machine learning-based technique called chained quantile regression and a stochastic method explicitly designed for discontinuous distributions. Its application reduces the data-simulation mismatch of the multivariate approach for identifying prompt photons to the percent-level. The impact of the associated systematic uncertainty in the Higgs to diphoton differential fiducial cross-section measurement is reduced from $5\%$ in its previous iteration to $1.5\%$. The on-detector electronic components of the barrel part of the CMS electromagnetic calorimeter will be replaced for the start of the High Luminosity LHC. However, some components will be kept in place to keep the upgrade time and cost-efficient. One of these components is the passive motherboard connecting the avalanche photodiodes, which collect the scintillating light from the crystals to the on-detector electronics. Thermal cycling interrupted with electric tests of the motherboard is performed to simulate the operating conditions until the end of the High Luminosity LHC. The reliability analysis shows that no more than $1.8\%$ of the channels can show an increase in electronic noise at a confidence level of $95\%$. The upper limit for channels with a total failure is predicted to $0.5\%$ of all channels of the barrel part of the CMS electromagnetic calorimeter at $95\%$ confidence level, which matches the targeted reliability