25th International Workshop on Next Generation Nucleon Decay and Neutrino Detectors (NNN26)

Charged-Particle Topology Reconstruction with an In-Liquid SiPM Array in CANDY

Not scheduled

20m

Exhibition Hall

Poster Poster Session

Speaker

MinSu Kwak

Description

We present charged-particle topology reconstruction with CANDY, a liquid scintillator detector containing a sparse three-dimensional SiPM array. A simulation-trained, time-informed convolutional neural network achieves median entry- and exit-point residuals of 1.91 cm and 2.39 cm for simulated top-to-bottom through-going muons. Reconstructed endpoints in cosmic-ray muon data are geometrically consistent with external trigger-counter acceptance regions. The framework also reconstructs simulated positron production vertices with a median residual of approximately 4.5 cm. These results demonstrate the potential of in-liquid SiPM arrays for topology-sensitive reconstruction in future liquid scintillator neutrino detectors.