An Efficient MPI‐Based Parallel Implementation of WCS‐FDTD Method

ABSTRACT

A parallel weakly conditionally stable finite‐difference time‐domain (WCS‐FDTD) method based on the message passing interface is proposed. It enables efficient simulation of large‐scale problems with fine structures in two directions. By employing a flexible domain decomposition strategy and efficient inter‐process communication, a scalable parallel implementation of the WCS‐FDTD method is realized. Furthermore, redundant coefficient matrices are eliminated according to the target characteristics and further compressed, significantly reducing the memory consumption. The Thomas algorithm is incorporated into the implicit field‐updating procedure, leading to additional acceleration of the computation. Numerical results demonstrate that the proposed method maintains good accuracy while achieving a significant reduction in memory consumption and an improvement in computational performance, achieving a parallel efficiency of approximately 90% when using 32 processors.