Wuppertal 2015 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 62: Trigger 1
T 62.7: Talk
Wednesday, March 11, 2015, 18:15–18:30, G.10.06 (HS 6)
A Neural Network z-Vertex Trigger for Belle II — •Sara Neuhaus1, Sebastian Skambraks1, Fernando Abudinen2, Yang Chen1, and Christian Kiesling2 for the Belle II collaboration — 1Technische Universität München — 2Max-Planck-Institut für Physik, München
In the Belle II experiment the efficiency of the track trigger could be increased by reconstructing the z-coordinate of track vertices at the first trigger level and rejecting tracks not coming from the interaction region, which form a large part of the machine background. The presented method employs neural networks to estimate the z-vertex without explicit track reconstruction. Input data is taken from the central drift chamber, using both the wire coordinates and the drift times for each hit. Neural networks are general function approximators that can learn nonlinear dependencies from real data without the need of an explicit model. However, using a priori knowledge about the track in a meaningful way can help to train more efficient networks, in terms of both prediction quality and network size. Such input information is provided by the Belle II 2D track trigger and is used explicitly in the calculation of the input values for the neural network. The algorithms for the input representation will be presented together with estimations for the trigger efficiency and the rejection capability.