Non Overlapping Domain Decomposition Method In Electromagnetic Scattering Analysis

More and more attention is paid to fast and accurate numerical simulations in the computational electromagnetism. The integral equation method is one of the most popular methods because of its high accuracy. However, the integral equation method costs a lot of computing resources. It is difficult to analyze the electrically large objects with the integral equation method.In this thesis, the non-overlapping domain decomposition method based on the integral equation(IE-DDM) is introduced to reduce the memory cost. The object is divided into few sub-domains, which are extended to closed structures and solved by recycling Krylov subspace method with the integral equation. A few outer Krylov iterations are needed to obtain the precise scattering field. Only the sub-domains’ information and a little global information is stored to compute each sub-domain by recycling Krylov subspace method.Combined IE-DDM with MLFMA, IE-DDM-MLFMA iterative algorithm is obtained to analyze the electrically-large target. MLFMA is used to accelerate the speed of the computed-sub-domain and the interaction between the incident-sub-domain and the computed-sub-domain in the IE-DDM-MLFMA iterative algorithm. The numerical examples show that IE-DDM-MLFMA reduces greatly the memory cost. Parallel technology is used in each sub-domain to reduce the total compute time, which is called the IE-DDM-MLFMA-Parallel algorithm. All in all, less computation time and memory are needed for the electrically-large objects.