Electron reconstruction with the ATLAS inner detector

The LHC will produce proton-proton collisions at a centre-of-mass energy of 14 TeV . ATLAS is a general-purpose detector for the LHC, sensitive to a wide range of physics processes. The total LHC production cross-section is dominated by QCD and so the search for rare physics events requires that ATLAS is able to reconstruct and identify leptons with high efficiency and accuracy. Electrons are measured in two ATLAS sub-detectors: the inner tracker reconstructs the trajectories of all charged particles, providing an estimate of the momentum; while the energies of electrons are measured by the electromagnetic calorimeter. The baseline track fitter for the inner detector is the Kalman filter (KF). The KF is optimal only when all measurement errors and material interactions can be described by gaussian probability density functions. Electrons loose energy in matter predominantly through bremsstrahlung, which is a strongly non-gaussian process described by the Bethe-Heitler distribution. In this case, a non-linear fitter may provide a better estimate of the trajectory of an electron than the KF. A non-linear generalisation of the KF, the gaussian-sum filter (GSF), has been developed and validated. The Bethe-Heitler distribution is approximated as a weighted sum of gaussian components. The performance of the GSF has been compared to the KF using both the simulated response of the detector and real data taken at the 2004 ATLAS test-beam. The performance of the GSF in simulation was first studied using samples of single electron events. The momentum resolution obtained using the GSF is, in general, better than the KF. Several LHC physics processes have also been simulated. Invariant mass distributions, obtained using the GSF, are superior to those from the KF in most cases. When the GSF was used to reconstruct real data from the test-beam, the momentum resolution was found to be much worse than predicted by simulation. Two factors contribute to the large discrepancy: the residual misalignment of the detector elements; and the material in the upstream beam-line. The impact of upstream material on the momentum resolution has been investigated in detail. The amount of material in the beam-line must be known precisely, both to quantify the performance of the track fitters and for the correct calibration of the electromagnetic calorimeter.