Track Reconstruction with Graph Neural Networks on a General Purpose Particle Collider Experiment

Abstract

In this paper we discuss the theoretical motivations behind building particle colliders. We then briefly look at the CMS detector as an example of a modern particle detector and study how it works. Finally, we reconstruct particle tracks from simulated CMS detector data using graph neural networks. We train various models at different transverse momenta thresholds, $p_T^{min}$, and assess how accurately the graph neural networks are able to reconstruct the tracks made by the particles as they interact with the detector.