Generative AI for Particle Shower Simulation

Particle interactions are governed by the Standard Model, but accurate modeling of those interactions in realistic media still depends on large Monte Carlo campaigns tuned to laboratory measurements. These simulations become especially time- and resource-intensive at the highest energies. We present challenges in particle astrophysics experiments, focusing on the IceCube Neutrino Observatory, where modeling hadronic air showers in the atmosphere and secondary cascades from deep inelastic scattering in ice is critical. We explore the use of generative AI to synthesize particle showers under these conditions as an alternative to full Monte Carlo generation.