GPU-Accelerated HO-SIE-DDM Using NVIDIA CUDA for Analysis of Multiscale Problems

In this paper, a novel non-overlapping and nonconformal domain decomposition method (DDM) based on surface integral equation (SIE) is proposed for the electromagnetic scattering analysis of multiscale PEC objects. In this SIE-DDM, conformal/nonconformal meshes can be achieved by adopting an interior penalty term. Besides, higher order hierarchical vector (HOHV) basis functions defined on the second-order curved triangular patches are used to reduce computational time and memory requirements. In order to further enhance the efficiency of modeling subdomains in this HO-SIE-DDM, the graphic processing unit (GPU) parallel technology on Compute Unified Device Architecture (CUDA) platform is also introduced to accelerate the matrix equation solution. Numerical results are given to demonstrate the validity of this HO-SIE-DDM with GPU-CUDA.