Fast pattern recognition with the ATLAS L1 track trigger for the HL-LHC

A fast hardware based track trigger for high luminosity upgrade of the Large Hadron Collider (HL- LHC) is being developed in ATLAS. The goal is to achieve trigger levels in high pileup collisions that are similar or even better than those achieved at low pile-up running of LHC by adding tracking information to the ATLAS hardware trigger which is currently based on information from calorimeters and muon trigger chambers only. Two methods for fast pattern recognition are investigated. The first is based on matching tracker hits to pattern banks of simulated high momentum tracks which are stored in a custom made Associative Memory (AM) ASIC. The second is based on the Hough transform where detector hits are transformed into 2D Hough space with one variable related to track pt and one to track direction. Hits found by pattern recognition will be sent to a track fitting step which calculates the track parameters . The speed and precision of the track fitting depends on the quality of the hits selected by the pattern recognition step. The figure of merit of the pattern recognition is measured by the efficiency for finding hits from high $p_\text{T}$ tracks and the power of rejecting hits from low pT tracks and fakes. We will describe the implementation of the two methods for Silicon Tracker (SIT) for ATLAS HL- LHC upgrad and compare the performance using full event simulation with 200 event pile-up .