Efficient Performance of MLSSM-MLFMA Algorithm Using Adaptive Grouping Technique for Electromagnetic Problems

In this communication, an adaptive grouping technique-based multilevel simply sparse method multilevel fast multipole algorithm (MLSSM-MLFMA) scheme is proposed to analyze objects with local fine structures efficiently. The MLSSM-MLFMA algorithm utilizes MLFMA to accelerate dealing with the boxes whose size is larger than 0.2 wavelength and uses MLSSM to decompose the boxes whose size is less than 0.2 wavelength. Meanwhile, the adaptive grouping technique is applied to improve the performance of MLSSM-MLFMA algorithm. The conventional octree-based grouping scheme divides the target into boxes with the same size at each level. It leads to some boxes that contain a large number of discrete unknowns, while some boxes contain fewer unknowns to simulate the object with local fine structures. Therefore, the computational efficiency of MLSSM-MLFMA algorithm will be improved. By using the proposed grouping technique, the memory requirements and computational times are much less than those of conventional octree-based grouping scheme for analyzing the objects with local fine structures.