Machine Learning Application for Particle Physics: Mexico’s Involvement in the Hyper-Kamiokande Observatory

The Hyper-Kamiokande (Hyper-K) observatory, the successor to the Super-Kamiokande (Super-K) experiment, will be the largest underground water Cherenkov neutrino observatory in the world. Hyper-K will be utilized to observe high energy neutrinos coming from the Sun, supernovas, and a neutrino beam from the Japan Proton Accelerator Research Complex (J-PARC). Hyper-K’s ultimate goal is to measure neutrino properties accurately, leading to quantifying the associated Charge-Parity violation in the leptonic sector, and thus to enhance the current understanding of the matter-antimatter asymmetry in the universe. Due to Hyper-K’s construction beginning in 2020, nowadays it’s a suitable time to perform Monte-Carlo simulations and test different Machine Learning (ML) analysis techniques such as Convolutional Neural Networks (CNN) for prototype development, besides exploring different detector configurations. The present chapter describes the participation of Mexico in the Hyper-K observatory, focusing on how ML and supercomputing can be used to design sensors, like the ones found in multi-photomultiplier tube (mPMT) arrays, to be tested on experiments and Hyper-K prototypes. Preliminary results show that ML techniques are good at distinguishing between muon and electron neutrino candidates, displaying comparable results to those of the likelihood analysis used on Super-K.