Heidelberg 2022 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 53: Data Analysis, Information Technology and Artificial Intelligence 3
T 53.4: Talk
Tuesday, March 22, 2022, 17:00–17:15, T-H38
Progressive Generative Adversarial Networks for High Energy Physics Calorimeter Simulations — •Simon Schnake1 , 2, Kerstin Borras1, 2, Dirk Krücker1, Florian Rehm2, 3, and Sofia Vallecorsa3 — 1DESY, Hamburg, Germany — 2RWTH Aachen, Germany — 3CERN openlab, Geneva, Swiss
The simulation of particle showers in calorimeters is a computational demanding process. Deep generative models have been suggested to replace these computations. One of the complexities of this approach is the dimensionality of the data produced by high granularity calorimeters. One possible solution could be progressively growing the GAN to handle this dimensionality. In this study, electromagnetic showers of a (25x25x25) calorimeter in the energy range of 10 - 510 GeV are used to train generative adversarial networks. The resolution of the calorimeter data is increased while training. First results of this approach are shown.