GPU-based Computing of RF Propagation

ORAL

Abstract

Tech-X has applied its GPU computing framework to ICRF propagation. This framework uses finite-difference methods for both modeling the antennas and the linear plasma response. This required developing robust cut-cell meshing including the ability to heal bad surfaces. The curved surfaces of the perfect conductor antennas are modeled by the Dey-Mittra algorithm, which has now been ported to GPUs. The linear-plasma-dielectric algorithm has been improved using the method of triads introduced by Bauer et al [1], and the implementation is underway. This talk will demonstrate the workflow from problem setup to visualization as well as show the results of convergence tests.

*Work supported by the Department of Energy grant DE-SC0018319

Authors

  • John Cary

    • Tech-X Corporation and University of Colorado Boulder
    • University of Colorado, Boulder
    • University of Colorado
  • Carl Bauer

    • Tech-X Corporation
  • Marc Durant

    • Tech-X Corporation
  • Tom Jenkins

    • Tech-X Corp
    • Tech-X Corporation
  • D. N. Smithe

    • Tech-X Corp
    • Tech-X
    • Tech-X Corporation