Position reconstruction of S2 signals near the edges of the DARWIN detector

DARWIN is a planned next-generation dark matter detector that will use a 40-ton liquid xenon time projection chamber to search for WIMP dark matter, neutrinoless double-beta decay, axion-like particles, measure solar neutrino fluxes, and investigate other physics channels. It will be sensitive to WIMP in the mass range from 5 GeV to above 10 TeV with cross section down to the neutrino floor. Building such a sensitive detector requires probing various designs to identify which have the best performance. In this talk, different designs of the gas region of the DARWIN detector are investigated using Monte Carlo simulation and Convolutional Neural Networks (CNNs) in order to improve the position reconstruction of the electroluminescence (S2) signals near the edges of the detector. Events are generated using a novel, GPU-based simulation framework. Then, CNNs are deployed to compare accuracy of position reconstruction for several alternative designs. The talk will briefly introduce the DARWIN observatory and present methods and current results of the attempt to improve the S2 position reconstruction near the edges of the detector.