This repo contains work done with the Texas Tech group for the CMS collaboration at CERN, specifically in close collaboration with Dr. Federico De Guio. It is a brief study of the feasibility of using machine learning techniques, specifically shallow neural networks, in hadronic dual-readout calorimetry, which simultaneously measures scintillation light and Cherenkov light. The electromagnetic (em) shower fraction can be calculated by comparing these two signals on an event-by-event basis. We show that machine learning has the potential to allow both scintillation and Cerenkov to be read out from a single channel while still allowing for an accurate calculation of the em shower fraction. This short report validates single-channel dual-readout calorimetry as a feasible technique in future high energy particle detector calorimeters. Moreover, there are possible latency advantages if these neural networks are hardwired into FPGAs, which could possibly aid a Level 1 trigger.
See attached pdf file for more. Please note that this repo and attached draft are still in progress.