Search for the decay of a top quark superpartner into a top quark and a neutralino using multivariate techniques at the ATLAS experiment

This master thesis describes a search for Supersymmetry at the ATLAS experiment at the LHC. For this search it is necessary to know the differences between the analysed decay of the top quark superpartner (stop quark) into a top quark and a neutralino and the corresponding backgrounds. Therefore distributions with different behaviours for these decays are needed. Separating variables are for example the missing transverse energy, the transverse mass, the perpendicular component of the missing transverse energy to the leptonically decaying top quark and some angular distributions between the decay products of the top quarks, the missing transverse energy, or the leading jets. Some of these variables are also used in the official ATLAS search. These variables are used in a neural network which is trained to differentiate between the stop decay and the two main backgrounds – $t\bar{t}$ and W+jets. This new method was optimised in order to reach a better signal to background ratio than the simple cut analysis. The outcoming distribution does not show any significant excess in the comparison of data and the Standard Model prediction. Therefore the distribution is used to set limits on the stop and neutralino masses. The limits using the neural network are improved compared to those using simple cuts especially for the bulk and the diagonal region.