Flavor identification of atmospheric neutrinos in JUNO with machine learning

Abstract

The Jiangmen Underground Neutrino Observatory (JUNO) is designed to determine neutrino mass ordering (NMO) using a large liquid scintillator detector located in southern China. While JUNO's NMO sensitivity mostly comes from reactor neutrinos, atmospheric neutrino oscillations in JUNO can provide complimentary sensitivity via matter effects, and enhance its overall sensitivity in a joint analysis. Neutrino flavor identification is crucial to atmospheric neutrino oscillation measurements, but is traditionally a very difficult task in liquid scintillator detectors such as JUNO. In this paper, we present a novel method for the flavor identification of atmospheric neutrinos in JUNO with machine learning techniques. This method takes features from photomultiplier tube waveforms as inputs, and shows promising results with JUNO simulation. This method could also be applied to other liquid scintillator detectors, potentially benefiting other future atmospheric neutrino oscillation experiments.