Measurement of the $t\bar{t}b\bar{b}$ production cross-section with 8 TeV ATLAS data

his thesis presents the measurement of the $t\bar{t}b\bar{b}$ production cross-section, using a dataset of 20.3 fb$^{-1}$ of $pp$ collisions collected with the ATLAS detector at $\sqrt{s}=8$ TeV. The measurement is based on a cut-and-count method, using a sample of events with exactly four $b$-tagged jets, which is shown to have a high purity in signal events. The measurement exploits the most precise jet energy scale and $b$-tagging calibrations and is performed in a fiducial phase space that is designed to minimize the model dependence of the measurement. The fiducial cross-section is measured to be $\sigma_{t\bar{t}b\bar{b}}^{\mathrm{fid}}=18.9\pm 3.5\mathrm{\ (stat)} ^{+5.6}_{-5.5} \mathrm{\ (sys)}\pm 0.6 \mathrm{\ (Lumi)}$ fb, or subtracting the contribution from $t\bar{t}H(b\bar{b})$ and $t\bar{t}Z(b\bar{b})$ final states, $\sigma_{t\bar{t}b\bar{b}}^{\mathrm{fid},\mathrm{QCD}}=17.8\pm 3.5\mathrm{\ (stat)} ^{+5.9}_{-5.7} \mathrm{\ (sys)}\pm 0.6\mathrm{\ (Lumi)}$ fb. The result is compared with a multitude of theoretical predictions, including different NLO calculations matched to a parton shower, which constitute the most precise predictions available to date, as well as with a series of models that differ in the description of the $g\rightarrow b\bar{b}$ splitting. It is shown that the most extreme $g\rightarrow b\bar{b}$ splitting model overestimates the observed rate of $t\bar{t}b\bar{b}$ production and that the measurement favors calculations performed with renormalization/factorization scales which are softer than the scales usually employed in similar calculations.