Real-time Machine Learning classifier to identify long-lived particles at the Large Hadron Collider.

The Large Hadron Collider at CERN provides the highest-ever proton-proton particle collisions achieved in the laboratory. A major goal of the LHC is to discover physics beyond the Standard Model. This work contributes to an upgrade of the hardware-based data acquisition system for the CMS experiment aimed at detecting new long-lived particles. In contribution to the effort to properly reconstruct particles displaced from the proton-proton collision point, a machine learning classifier was developed to distinguish signal particles that are correctly reconstructed with displacement from background particles that originate from the collision point but are incorrectly reconstructed as displaced. A boosted decision tree was used to classify the quality of displaced particle reconstruction, and in-depth investigation into the characteristics of these tracks was used to optimize the classifier.