CaloPointFlow - Generating Calorimeter Showers as Point Clouds

Journal of Physics Conference Series, vol. 3206, pp. 012108

Abstract

Abstract Precise simulations are essential for advancing particle physics research, but they require high computational resources due to the complexity of calorimeter interactions. To reduce this cost, we propose a novel generative model that interprets calorimeter showers as point clouds rather than 3D images. Each hit is modelled as part of a hit distribution depending on a global latent calorimeter shower distribution. Our model is based on PointFlow (Yang et al., 2019) and consists of a permutation invariant encoder and two normalizing flows: one for modelling the global latent distribution and another for modelling individual hits conditioned on it. We present first results and compare them with state-of-the-art voxel methods.

Authors 6

1. Simon Schnake corresponding Aachen

   Deutsches Elektronen-Synchrotron DESY · RWTH Aachen University

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, 22607 Hamburg,

   RWTH Aachen University

2. Benno Käch

   Deutsches Elektronen-Synchrotron DESY

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, 22607 Hamburg,

3. M. Scham Aachen

   Deutsches Elektronen-Synchrotron DESY · RWTH Aachen University

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, 22607 Hamburg,

   RWTH Aachen University

4. Dirk Krücker Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University

5. K. Borras Aachen

   Deutsches Elektronen-Synchrotron DESY · RWTH Aachen University

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, 22607 Hamburg,

   RWTH Aachen University

6. Sofia Vallecorsa

   European Organization for Nuclear Research

   Affiliation as printed

   CERN, 1211 Geneva 23,