Reconstruction of electrons with the Gaussian-Sum filter in the ATLAS inner detector

The ATLAS detector's ability to reconstruct and identify electrons with high precision is critical both for the calibration of the detector and for the analysis of many physics processes. High energy electrons traversing matter loose energy predominantly by bremsstrahlung. A well known model for bremsstrahlung is the Bethe-Heitler distribution, which is highly non-Gaussian. Linear fitters, such as the Kaiman filter (KF), model energy losses as Gaussian distributions and thus are not optimal for the reconstruction of electrons. The Gaussian-Sum filter (GSF) is a non-linear generalisation of the KF. This algorithm has been implemented for the reconstruction of electrons within ATLAS. The GSF models the Bethe-Heitler distribution by a weighted sum of Gaussian distributions. A comparison between the reconstruction performance of the GSF and the KF is presented using simulated events generated using the geometry of the 2004 ATLAS test-beam.