Searching for Supersymmetry with the ATLAS detector

This thesis presents a new method by which one may use data from the ATLAS detector of the Large Hadron Collider at CERN to measure the parameters of the theory of supersymmetry (SUSY). The technique uses a Markov Chain Monte Carlo sampling algorithm to combine measurements of exclusive variables, in the form of kinematic endpoints that arise in the invariant mass distributions of leptons and jets given out in sparticle decay chains, with inclusive data, in the form of the cross-section of events passing a missing transverse energy cut. This improves the precision of sparticle mass measurements (beyond that obtained using exclusive data alone), whilst also enabling experimental uncertainties to be handled in an intuitive fashion. The method is demonstrated on an mSUGRA benchmark model, and is also used to constrain a model with a greater number of parameters. Throughout, an attempt is made to break some of the unrealistic assumptions that characterise current SUSY search techniques, and to this end it is successfully demonstrated that one can use the Markov Chain method to obtain precise results even if it is not possible to precisely determine which sparticles produced the endpoints in the invariant mass distributions. This decay chain ambiguity is extended by looking at a sample SUSY model with non-universal Higgs masses, in which cascade decays featuring three body decays become more prominent. The positions of the associat ed endpoints are calculated for a squark cascade decay, and are subsequently studied with the aid of a benchmark model. In addition, this thesis presents work relating to the development of an online monitoring package for the ATLAS Semi-Conductor Tracker (SCT). A tool for the calculation of noise occupancies is developed and tested against existing calibration data, prior to its application to various data sets obtained during the SR1 cosmic commissioning tests at CERN in the summer of 2006. It is found that the modules of the SCT included in the test have an average noise occupancy of less than $5 \times 10^{-4}$ per strip per event, and thus meet the design specification.