Novel XENON1T Event Localization Using Graph Neural Networks

Novel application of a Graph Neural Network for event localization in the XENON1T dark matter direct detection time projection chamber. Our graph neural network made use of the signals produced by events in XENON1T to reconstruct the positions of said events. This is an early application of a graph neural network in particle physics and for regression problems. We have shown that such a neural network is competitive with the state-of-the-art for this type of problem.