Fast track seed selection for track following in the Inner Detector Trigger track reconstruction: Proceedings

During ATLAS Run 2, in the online track reconstruction algorithm of the Inner Detector, a large proportion of the CPU time was dedicated to the track finding. With the proposed HL-LHC upgrade, where the event pile-up is predicted to reach $\langle \mu \rangle=200$, track finding will see a further large increase in CPU usage. Moreover, only a small subset of track candidate seeds is accepted after the track finding procedure, spending the CPU time on seeds that are discarded. Therefore, a computationally cheap track candidate seed pre-selection procedure based on approximate track following was designed, which is described in this report. The algorithm uses a simplified track extrapolation and a combinatorial Kalman filter simplified by a reference-related coordinate system to find the best track candidates. For such candidates, a set of numerical features were created to classify seeds using a Support Vector Classifier, tuned with a high True Positive rate to ensure no significant loss of track finding efficiency. The algorithm was implemented into the Athena framework for online seed pre-selection, and could be potentially adapted for the ITk geometry for the Run 4 of the HL-LHC.