Design Study in Ultra-High Energy Neutrino Reconstruction Performance by Convolutional Neural Network with In-Ice Radio Array.

Observing radio signals from Askaryan Effect produced by showers of particles when ultra-high energy neutrinos interact in glacial ice is the fundamental concept of existing experiments like Askaryan Radio Array (ARA) at South Pole and Radio Neutrino Observatory Greenland (RNO-G). A convolutional neural network is used to study vertex and neutrino reconstruction by an array of only vertical dipole antennas. The vertical and horizontal dimensions of the array are varied to optimize performance. The neural network uses true travel times and amplitudes of arriving signals obtained by simulation using NuRadioMC module in Python using South Pole ice model as the results could provide valuable experience toward IceCube Gen2 Radio.