% IMPORTANT: The following is UTF-8 encoded.  This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.

@INPROCEEDINGS{Buhmann:600309,
      author       = {Buhmann, Erik and Eren, Engin and Gaede, Frank and
                      Kasieczka, Gregor and Korol, Anatolii and Korcari, William
                      and Krüger, Katja and McKeown, Peter},
      title        = {{G}enerative {M}odeling with {D}iffusion {N}eural
                      {N}etworks for {F}ast {S}imulation of {E}lectromagnetic
                      {S}howers in the {I}nternational {L}arge {D}etector},
      school       = {Dresden University of Technology},
      reportid     = {PUBDB-2023-07877},
      year         = {2023},
      abstract     = {In high energy physics, detailed and time-consuming
                      simulations are used for particle interactions with
                      detectors. For future experiments and the upcoming
                      High-Luminosity phase of the Large Hadron Collider (HL-LHC),
                      the computational costs of conventional simulation tools are
                      expected to exceed the projected computational
                      resources.Generative neural networks (GNNs) have the
                      potential to provide a fast and accurate alternative. So far
                      most of the studies of GNNs for fast simulations have used
                      data represented in the form of a regular grid since it is
                      possible to apply modern machine learning algorithms from
                      image processing that are well optimized and developed.In
                      fast simulations with GNNs, it is crucial to be able to
                      place GNNs into the simulation pipeline, and since many of
                      today's detector systems are not regular in terms of the
                      positions of the active cells, it is very hard to represent
                      the data in a form suitable for training the GNN.This work
                      focuses on the development of a GNN for speeding up the
                      simulation of electromagnetic showers in the electromagnetic
                      calorimeter of the International Large Detector (ILD). In
                      particular, a Diffusion Model is trained on Geant4 steps,
                      where the electromagnetic shower is presented as a 3D point
                      cloud to avoid the irregularities of the detector geometry
                      and thereby generate showers anywhere in the calorimeter.},
      month         = {Mar},
      date          = {2023-03-20},
      organization  = {86. Annual Meeting of DPG and DPG
                       Spring Meeting of the Matter and Cosmos
                       Section , Dresden (Germany), 20 Mar
                       2023 - 24 Mar 2023},
      subtyp        = {Other},
      cin          = {FTX},
      cid          = {I:(DE-H253)FTX-20210408},
      pnm          = {623 - Data Management and Analysis (POF4-623) / AIDAinnova
                      - Advancement and Innovation for Detectors at Accelerators
                      (101004761) / 05D23GU4 - Verbundprojekt 05D2022 - KISS:
                      Künstliche Intelligenz zur schnellen Simulation von
                      wissenschaftlichen Daten. Teilprojekt 1. (BMBF-05D23GU4)},
      pid          = {G:(DE-HGF)POF4-623 / G:(EU-Grant)101004761 /
                      G:(DE-Ds200)BMBF-05D23GU4},
      experiment   = {EXP:(DE-MLZ)NOSPEC-20140101},
      typ          = {PUB:(DE-HGF)6},
      url          = {https://bib-pubdb1.desy.de/record/600309},
}