Machine-learning aided detector optimization of the Pacific Ocean Neutrino Experiment

The Pacific Ocean Neutrino Experiment (P-ONE) is a planned cubic-kilometer-scale neutrino detector in the Pacific Ocean. P-ONE will measure high-energy astrophysical neutrinos to characterize the nature of astrophysical accelerators. Using existing deep-sea infrastructure provided by Ocean Networks Canada (ONC), P-ONE will instrument the ocean with optical modules - which host PMTs and readout electronics - deployed on several vertical cables of about 1 km in length. While the hardware design of the first prototype cable is currently being finalized, the detector geometry of the final instrument (up to 70 cables) is not yet fixed. Conventional design optimization typically requires extensive Monte-Carlo simulations, which limits the testable search space to a few configurations. In this contribution, we present the progress of optimizing the detector design using machine-learning-based surrogate models, which replace the computationally expensive MC simulations. By providing gradients, these models also allow for the efficient computation of detector resolutions via the Fisher Information Matrix, without having to rely on specific event-reconstruction algorithms.