ICSC and Spoke 2 - Where Are We Now?

FaDER - Assisting real-time track reconstruction for LHC experiments

Not scheduled

20m

Conference Room (Physics Dept and INFN, Catania)

Open Calls

Speaker

Alessandro Zaio (Istituto Nazionale di Fisica Nucleare)

Description

The real-time track reconstruction task for LHC experiments shows a processing time which increases significantly as a function of the average number of proton-proton collisions per bunch crossing. The future upgrade to the High-Luminosity LHC (HL-LHC), with way higher levels of simultaneous collisions, could thus lead to a considerable growth in computational cost for the current trigger algorithms. To face this issue, a machine-learning-based technique to assist tracking by filtering out background hits is presented and characterized, as part of the FaDER project. The algorithm is based on a Convolutional Neural Network architecture, to target final deployment on FPGA boards.