A Neural-Network Clusterisation Algorithm for the ATLAS Silicon Pixel Detector

A novel technique using a set of artificial neural networks to identify and split merged measurements created by multiple charged particles in the ATLAS pixel detector is presented. Such merged measurements are a common feature of boosted physics objects such as tau leptons or strongly energetic jets where particles are highly collimated. The neural networks are trained using Monte Carlo samples produced with a detailed detector simulation. The performance of the splitting technique is quantified using LHC data collected by the ATLAS detector and Monte Carlo simulation. The number of shared hits per track is significantly reduced, particularly in boosted systems, which increases the reconstruction efficiency and quality. The improved position and error estimates of the measurements lead to a sizable improvement of the track and vertex resolution.