Looking for the Charged Higgs Boson - Simulation Studies for the ATLAS Experiment

The discovery of a charged Higgs boson (H+) would be an unambiguous sign of physics beyond the Standard Model. This thesis describes preparations for the H+ search with the ATLAS experiment at the Large Hadron Collider at CERN. The H+ discovery potential is evaluated, and tools for H+ searches are developed and refined. The $H^{+} \to \tau \nu$ decay mode has been known as the most promising H+ discovery channel. Within this thesis, first studies of this channel with realistic detector simulation, trigger simulation and consideration of all dominant systematic uncertainties have been performed. Although, as shown by these studies, the discovery sensitivity is significantly degraded compared to studies using a parametrized detector simulation, this channel remains the most powerful ATLAS H+ discovery mode. Future searches will rely on multivariate analysis techniques like the Iterative Discriminant Analysis (IDA) method. First studies indicate that a significant sensitivity increase can be achieved compared to studies based on sequential cuts. The largest uncertainty in H+ searches is the expected $t \overline{t}$ background contribution. It is shown that numbers obtained from simulated events could be off by a factor of two, decreasing the discovery sensitivity dramatically. In this thesis, the Embedding Method for data-driven background estimation is presented. By replacing the muon signature in $t \overline{t}$ events with a simulated $\tau$, events which allow an estimation of the background contribution at the 10% level are obtained. The ATLAS $\tau$ identification focuses on comparably clean environments like Z and W decays. To optimize the performance in high-multiplicity events like $H^{+} \to \tau \nu$ tau leptons are studied in $t \overline{t}$ and pile-up events. Variables which do not show discrimination power in highmultiplicity events are identified, and in some cases similar, more powerful variables are found. This allows to recover some of the performance loss and to increase the robustness of the $\tau$ identification. For the analysis of large amounts of data produced by the ATLAS detector, seamless interoperability of the various Grid flavors is required. This thesis introduces translators to overcome differences in the information system between a number of Grid projects, and highlights important areas for future standardization.