Karlsruhe 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 41: Trigger+DAQ 1

T 41.3: Talk

Tuesday, March 5, 2024, 16:30–16:45, Geb. 30.23: 3/1

Trigger Algorithm for Electron Identification with Neural Networks and Realization in Firmware — •Moritz Vogt, Dennis Layh, and Stefan Tapprogge — Institute for Physics, Mainz, Germany

As the LHC is upgraded to the High-Luminosity LHC, the instantaneous luminosity will increase significantly. To cope with the additional pile-up and the increase in the rate of background events, the triggering algorithms need to be improved. Following the High-Luminosity upgrade, the new forward Feature Extractor (fFEX) first level trigger module will have real-time access to ATLAS forward (|η|>2.5) calorimeter information with full granularity. The accurate identification of electrons in this region should allow a refined measurement of the weak mixing angle using the forward-backward asymmetry of the Z-boson decay.

This contribution will focus on the identification of electrons using neural networks in simulated events of the detector response in the Liquid Argon Endcap Calorimeters of the ATLAS detector (2.5<|η|<3.2). Since the trigger module will be realized with Field Programmable Gate Arrays (FPGA), the optimization of neural networks for deployment in firmware on these components is of utmost importance. Achieving a balance between signal efficiency, background rejection, resource utilization, and latency is the ultimate goal. This talk will present the results of the studies and give an outlook on future extensions.

Keywords: ATLAS; Neural Networks; FPGA; first level trigger