Speaker
Description
We present a GPU-accelerated optical simulation approach based on Simphony (sic), integrating the GeGeDe and edep-sim packages with Simphony through a thin plugin interface. Simphony, based on NVIDIA OptiX and developed as a fork of Opticks, performs GPU-accelerated optical-photon transport, including scattering, absorption, and detection. As part of this work, Simphony has been extended with GPU-based wavelength-shifting physics and asynchronous task management, enabling parallelism between CPU and GPU and improving overall throughput. Monte Carlo truth information is preserved throughout the GPU simulation, allowing detected photons to be traced back to their parent particles.
Using a DUNE-sized benchmark toy detector, we demonstrate agreement with Geant4, with less than 0.25% disagreement in detected photon counts and approximately 1000× speed-up for optical simulation. This GPU-based Simphony workflow enables efficient simulation of large numbers of optical photons and complex optical systems, making detailed optical studies more practical for large-scale detectors and novel detector concepts. The modular approach can be readily extended to other scintillation-based detector technologies.