Modeling and measuring the $t\bar{t}\gamma$ process with the ATLAS detector at 8 TeV

The Monte Carlo generators for the $t\bar{t}\gamma$ process are available in leading order and next-to-leading order. Using a next-to-leading order generator features a better precision. Due to a large amount of possible diagrams for the $t\bar{t}\gamma$ process however, the next-to-leading order generators are not able to accurately describe photons radiated from the decay products of the top quark. Leading order generators on the other hand are able to have photons in the final state. This includes interference effects of initial and final states, which cannot be described by next-to-leading order generators. With the addition of a QED generator, to simulate the final state photons, the modeling of leading order and next-to-leading-order generators are studied. The cross section of $t\bar{t}\gamma$ production in collisions of protons is predicted by the Standard Model and has to be measured by experiment. The measurement presented in this thesis is done using $t\bar{t}\gamma$ events in the dilepton channel. Dileptonic events were not used for the $t\bar{t}\gamma$ process in general yet, because of the small branching fraction. The analysis uses events from the e$\mu$-channel only, because of the very high signal-to-background ratio, which is not blurred by background processes involving the $Z$ boson. A cut-and-count method is used for the measurement. To get the optimal set of requirements for the measurement, the event selection was optimized by minimizing the relative uncertainty on the expected cross section.