ArborTracking: a tree topology track pattern recognition algorithm

Abstract

The purpose of this work is to develop a novel pattern tracking algorithm to be used on the detectors of the future electron–positron colliders. ArborTracking, a light-weighted tracking algorithm, has been developed based on the tree topology of track clusters and applied to the baseline detector of the circular electron–positron collider (CEPC). The algorithm collects all the hits in the tracker as a tree (forest), splits the tree branches to form the track segments, and merges the track segments to form the tracks. Compared with the general track following method, the algorithm has the advantages of low coding complicity and low CPU cost. The performances at different benchmarks are studied. The results are exhaustively listed showing that the method is approaching the limit of the detector. The tracking efficiencies on single muon sample and three-prong sample are both higher than 99%. The transverse momentum resolution reaches 0.1% level and the boson mass resolution reaches 4.7 GeV/c. The performances are similar with those of the baseline tracking algorithm of CEPC, and the physics requirement of CEPC is satisfied. The new tree pattern recognition algorithm is a necessary part in the CEPC software. And it is also a competitive algorithm on the market, which can be chosen by the future experiments.