Efficient Characteristic Mode Analysis Based on Data-Sparse Matrix Algorithm
Ting Wan, W. J. Wang, Y. Bao, Y. F. Chen · IEEE Transactions on Antennas and Propagation · 2024
The implicitly restarted Arnoldi method (IRAM), which relies on Krylov subspace iteration, is an effective approach for extracting desired partial eigenpairs in characteristic mode analysis (CMA). The primary limitation of IRAM is that the matrix–vector products (MVPs) of a dense matrix and its inverse need to be performed at each iteration step. This communication develops an efficient method to break through this limitation by incorporating a data-sparse matrix algorithm into the IRAM. The proposed method accelerates the MVP of a dense matrix in a data-sparse way and provides a data-sparse formatted lower-upper (LU) triangular decomposition algorithm to deal with the MVP of the inverse of a dense matrix. Numerical examples demonstrate the accuracy and efficiency of the proposed method.