GPU-Accelerated Optical Physics in Celeritas

ORAL

Abstract

Celeritas is a Monte-Carlo based particle transport code developed for computationally intensive applications in the energy and intensity frontier experiments. With the upgrade of the High Luminosity Large Hadron Collider (HL-LHC) and the increasing scale of neutrino and rare event detectors, the demand for high-fidelity, large-scalesimulations is increasing. Celeritas addresses these challenges through GPU-optimization and platform portability, providing electromagnetic physics and is currently extending its capabilities to optical simulation, with seamless integration into existing Geant4 applications by offloading particles to Celeritas. In this work, we present the current status, focusing on GPU optical photon simulation. To demonstrate performance gains, we perform benchmark comparisons of optical photon generation and transport between Celeritas and Geant4 for different experimental setups, positioning Celeritas as a scalable solution for the next-generation experiments.

*This work was supported by the DOE HEP Center for Computational Excellence at Oak Ridge National Laboratory under B&R KA2401022.

Presenters

  • Rashika Gupta

    • Oak Ridge National Laboratory

Authors

  • Rashika Gupta

    • Oak Ridge National Laboratory
  • Stefano C. Tognini

    • Oak Ridge National Laboratory
  • Seth R Johnson

    • Oak Ridge National Laboratory
  • Amanda L Lund

    • Argonne National Laboratory
  • Benjamin Morgan

    • University of Warwick
  • Elliott Biondo

    • Oak Ridge National Laboratory
  • Lima Guilherme

    • Fermi National Accelerator Laboratory
  • Hayden Hollenbeck

    • University of Virginia
  • Julien Esseiva

    • Lawrence Berkeley National Laboratory
  • Sakib Rahman

    • Brookhaven National Laboratory
  • Soon Yung Jun

    • Fermi National Accelerator Laboratory