Karlsruhe 2024 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 94: Trigger+DAQ 3
T 94.8: Talk
Thursday, March 7, 2024, 17:45–18:00, Geb. 30.23: 3/1
Upgrade of the Neural Network Track Trigger for Belle II — •Simon Hiesl1,2, Christian Kiesling1,2, Timo Forsthofer1,2, Kai Unger3, and Sebastian Skambraks1 — 1Max-Planck-Institute for Physics Munich — 2Ludwig Maximilian University of Munich — 3Karlsruhe Institute of Technology
The Single Track Trigger of Belle II is a Level 1 Neural Network Trigger implemented on hardware (FPGA boards).
This system is used to determine the vertex of charged tracks in order to exclude the dominant background from beam-gas interactions.
Candidate tracks are found by a novel 3D-Hough transformation algorithm which was previously only done in 2D. The new track model avoids fake tracks from background and suppresses tracks from outside of the interaction region. Using the new 3D track candidates, a set of neural networks predicts the track vertex along the beam axis and its polar emission angle.
The new method is expected to strongly improve the performance of the neural network trigger under harsh background conditions coming along with high luminosity.
Keywords: Belle II; Track Trigger; 3D-Hough Transformation; Neural Network; FPGA