A Machine Learning Reconstruction Method for Atmospheric Neutrino's Interaction Vertex and Muon Range in the JUNO Detector

Neutrino mass ordering (NMO) is one of the biggest remaining problems in particle physics. The Jiangmen Underground Neutrino Observatory (JUNO) is designed to solve this problem with a 20-kton liquid scintillator detector. Atmospheric neutrino oscillation measurement in JUNO offers independent sensitivity to NMO via matter effect, complimentary to its reactor neutrino measurement, which can greatly increase JUNO’s total sensitivity to NMO when combined. In this poster, we present a novel multi-purpose method for the reconstruction of atmospheric neutrino interaction’s detailed topological structure, including interaction vertex, muon range in νµ-cc interactions, etc., by extracting features from PMT waveforms and using them as inputs to machine learning models. This method reconstructs multiple objects within one framework, and has the potential to achieve better resolution than traditional methods. Preliminary performance estimations using Monte Carlo simulation are presented.