Implicit representation modeling neutrino event topologies in water-Cherenkov detectors

Neutrino experiments based on water-Cherenkov detectors have made significant contributions to our understanding of neutrino physics, but they face challenges in accurately modeling detector systematic parameters due to their large size necessary to overcome the smallness of the cross-section for weak interactions. While these experiments have achieved remarkable successes in the past, the future era of precision neutrino physics demands innovative techniques to better comprehend detector systematic uncertainties.

In this talk, I will present modern ideas to represent neutrino event topologies in preparation for the Hyper-Kamiokande experiment, a next-generation underground water-Cherenkov detector, poised to begin construction in the near future and serving as a far detector, positioned 295 km away, for a long baseline neutrino experiment utilizing the upgraded J-PARC beam in Japan. Moreover, it will have the capability to search for proton decay, atmospheric neutrinos, and neutrinos from astronomical sources with unprecedented sensitivity compared to its predecessor, Super-Kamiokande.

To address the challenges of modeling systematic uncertainties, I will explore the application of view rendering techniques, including Neural Radiance Field (NeRF), which can implicitly encode detector parameters such as water attenuation length and scattering. Furthermore, these techniques can be employed in event reconstruction, generating new events to enhance the understanding of systematic uncertainties. The ongoing work presented here aims to achieve computationally efficient and comprehensive treatment of systematic uncertainties by accelerating simulations and event reconstruction, enabling the variation of detector parameters for large water-Cherenkov detectors.