Speaker
Marcel Köpke
(Karlsruhe Institute of Technology)
Description
The time complexity of extensive air shower simulations rises approximately linearly with the incident particle energy for the CORSIKA 7 framework. The range of cosmic ray energies observed on earth covers several orders of magnitude. In order to simulate the highest energies in the cosmic ray spectrum, one has to introduce some sort of heuristic (e.g. thinning) which reduces runtime and preserves the shower properties to leading order. The physical content on higher order effects, like shower-to-shower fluctuations, is usually reduced. In this talk I am going to present my ideas on how to supplement current heuristics by training neural networks on CORSIKA simulations.
