CaloChallenge 2022: a community challenge for fast calorimeter simulation
Claudius Krause, M. Faucci Giannelli, G. Kasieczka, Benjamin Nachman, D. Salamani, David Shih, Anna Zaborowska, O. Amram, K. Borras, Matthew R. Buckley, Erik Buhmann, Thorsten Buss, Renato Cardoso, Anthony L. Caterini, N. Chernyavskaya, Federico Andrea Corchia, Jesse C. Cresswell, Sascha Diefenbacher, E. Dreyer, Vijay Ekambaram, Engin Eren, Florian Ernst, Luigi Favaro, M. Franchini, Frank Gaede, E. Gross, S.‐C. Hsu, Kristina Jaruskova, B. Kaech, Jayant Kalagnanam, R. Kansal, Taewoo Kim, D. Kobylianskii, Anatolii Korol, W. Korcari, D. Krücker, K. Krüger, Marco Letizia, S. Li, Q. Liu, X.T. Liu, Gabriel Loaiza-Ganem, T. Madula, Peter P. McKeown, I.-A. Melzer-Pellmann, V. M. Mikuni, Nam Nguyen, Ayodele Ore, Sofia Palacios Schweitzer, Ian Pang, Kevin Pedro, Tilman Plehn, Witold Pokorski, H. Qu, Piyush Raikwar, J. A. Raine, Humberto Reyes-González, L. Rinaldi, Brendan Leigh Ross, M. Scham, Simon Schnake, Chase Owen Shimmin, Eli Shlizerman, Nathalie Soybelman, Mudhakar Srivatsa, Kalliopi Tsolaki, S. Vallecorsa, Kyongmin Yeo, R. Zhang
Abstract
Abstract We present the results of the ‘Fast Calorimeter Simulation Challenge 2022’—the CaloChallenge. We study state-of-the-art generative models on four calorimeter shower datasets of increasing dimensionality, ranging from a few hundred voxels to a few tens of thousand voxels. The 31 individual submissions span a wide range of current popular generative architectures, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, diffusion models, and models based on conditional flow matching. We compare all submissions in terms of quality of generated calorimeter showers, as well as shower generation time and model size. To assess the quality we use a broad range of different metrics including differences in one-dimensional histograms of observables, KPD/FPD scores, AUCs of binary classifiers, and the log-posterior of a multiclass classifier. The results of the CaloChallenge provide the most complete and comprehensive survey of cutting-edge approaches to calorimeter fast simulation to date. In addition, our work provides a uniquely detailed perspective on the important problem of how to evaluate generative models. As such, the results presented here should be applicable for other domains that use generative AI and require fast and faithful generation of samples in a large phase space. Report Numbers : HEPHY-ML-24-05, FERMILAB-PUB-24-0728-CMS, TTK-24-43.