Electromagnetic shower identification in the SAND Calorimeter of the DUNE Near Detector

The Deep Underground Neutrino Experiment (DUNE) is a next generation, long-baseline accelerator experiment designed to significantly contribute to the study of neutrino oscillations with unprecedented level of sensitivity. It envisages to observe neutrinos from a high intensity wide-band neutrino beam with a Near Detector (ND) system at the Fermi National Accelerator Laboratory and a Far Detector (FD) at ∼ 1300 km from the beam source at Sanford Laboratory, in South Dakota. The System for on-Axis Neutrino Detection (SAND) is one of the three detectors of the ND complex which provides continuous on-axis beam monitoring and performs various physics measurements. This work concerns a study on the electromagnetic calorimeter (ECAL) capabilities of the SAND detector to perform e.m. showers and muon tracks discrimination. In particular, a new clustering algorithm to reconstruct events in the ECAL not exploiting at all Monte Carlo truth has been developed and studied for the first time. A multivariate analysis based on machine-learning techniques has been implemented to perform a classification of the clusters reconstructed by the clustering algorithm and study the capabilities in electromagnetic showers and muon tracks discrimination. The classification analysis is operated in two separated steps: (a) the multivariate methods are trained and validated using a dedicated dataset obtained by simulating particle guns of electrons and muons originated close to the ECAL; (b) the selected optimal classifiers are tested with simulated neutrino interactions in the SAND detector considering the case of a ν µ -dominated beam in neutrino mode. Finally, the calorimeter performance in discriminating electromagnetic showers from muons is preliminary evaluated in terms of the product of efficiency times the signal purity.