Synthetic Functions Expansion: Automation, Reuse, and Parallel

Abstract

This communication systematically summarizes the previous work of synthetic functions expansion (SFX) and three important features of this algorithm are discussed. First, for large-scale targets with complex geometries, two automatic schemes for geometric domain decomposition and defining of auxiliary sources are proposed which greatly decrease the manual workload in the preprocessing work. Second, for large-scale arrays, the synthetic functions (SFs) reusable scheme is proposed for both linear and nonlinear transformations. This scheme can improve the efficiency of constructing SFs and is especially suitable for large-scale arrays. Third, parallel algorithm is applied into SFX to further enhance its efficiency. As SFX deals with a large- scale electromagnetic system (EM system) block by block, parallel algorithm is an obvious and feasible way to improve its efficiency. Finally, a simplified ship and a cylindrical conformal array are adopted to verify the performance of parallel SFX. And results show that parallel SFX exhibits prominent advantages in memory cost compared with multilevel fast multipole method (MLFMM), while the efficiency and accuracy are comparable with the latter.