Fast and Efficient Algorithms for Computational Electromagnetics on GPU Architecture

Abstract

Magnetic diagnostics are essential for the operation and understanding of a magnetic confinement fusion (MCF) device, for example, by providing in real-time, the axisymmetric measurements needed for plasma equilibrium reconstructions or magnetohydrodynamic instabilities control. In this paper, we present a fast and robust implementation on GPU architecture of an accurate algorithm suitable for the computation of synthetic magnetic measurements used both in real time applications and off-line reconstruction procedures. The proposed procedure is first applied to benchmark problems, to validate the compute unified device architecture implementation, and then to cases of practical interests in MCF.