Investigation of the discovery potential for supersymmetry in Tau final states and measurement of the Tau identification efficiency for the ATLAS experiment

Despite its success, the Standard Model has a number of short-comings that lead particle physicists to believe the it is only a low-energy approximation of a more fundamental theory. One of the most promising candidates for an extension of the Standard Model is supersymmetry. From 2009 the search for Supersymmetry will be taken into a new energy regime with the Large Hadron Collider (LHC) experiments at CERN. A new inclusive search for SUSY in tau final states has been developed for the ATLAS experiment. The search focuses on the signature of taus, jets and missing transverse energy. Analyses with different jet multiplicities (4, 3 and 2-jets) have been studied. The requirement of the tau significantly reduces the abundant QCD multijet background making the mode potentially more robust than other modes already in use that focus only on jets and missing transverse energy as the signature. The discovery reach for R-parity conserving mSUGRA models has been studied for a centre-of-mass energy of 14 TeV and an integrated luminosity of 1fb-1. It has been shown that this tau mode is competitive with other inclusive search modes used by the ATLAS Collaboration, particularly for regions of the SUSY parameter space with high tan ß; where tau decays are enhanced. It has been shown that models with high tan ß and squark and gluino masses less than O(1 TeV) are within the discovery reach. Tau leptons will play an important role in the physics expected at the LHC both in Standard Model and beyond the Standard Model processes. However, due to their prompt decay, taus are challenging objects to identify, but the excellent tracking and calorimetry of the ATLAS detector should allow for efficient identification and reconstruction of hadronically decaying taus. The validation of the ATLAS tau identification will be important in early data. A new method for determining the tau identification efficiency using ttbar decays, which will be abundant at the LHC, has been developed. The method isolates the semileptonically decaying ttbar events by requiring high missing transverse energy, a number of high energy jets and by reconstructing the mass of the hadronically decaying t-quark. As a result, this method does not suffer from a large background from QCD multijet events and it does not require tau triggering. It also does not rely on b-tagging to reconstruct the top quark mass and is thus a suitable method for early data. The method has been tested using pseudo-data with reduced tau identification efficiencies and good agreement was found between the measured tau identification efficiencies and those expected. The statistical and systematic uncertainties on the tau identification efficiency measurement have been shown to be ±11% (±16%) and ±19% (±19%) respectively for taus with pT > 20 GeV (pT > 40 GeV) assuming an integrated luminosity of 1 fb-1 for a centre-of-mass energy of 14 TeV. Measurement of the tau identification efficiency in an early data taking phase with a centre-of-mass energy of 10 TeV using the new method has also been investigated. Despite this being a more challenging environment, it has been shown that the method for measuring the tau identification efficiency using ttbar events is able to correctly measure the tau identification efficiency for a centre of mass energy of 10 TeV and 200 pb-1 of integrated luminosity. The method is therefore a promising method for the 10 TeV data taking phase of the ATLAS experiment. The statistical and systematic uncertainties on the tau identification efficiency measurement have been shown to be ±28% and ±22% respectively in a worst case scenario where the systematic uncertainty on the SM background is 50%. For a systematic uncertainty of 30% the systematic uncertainty on the tau identification efficiency measurement is ±17%. The method appears to be a potentially promising method for measuring the tau identification efficiency in early data.