Modified adaptive cross approximation algorithm for analysis of electromagnetic problems

In order to efficiently analyze the large dense complex linear system arising from electric field integral equations (EFIE) formulation of electromagnetic scattering problems, the adaptive cross approximation (ACA) is applied to accelerate the matrix-vector multiplication operations. Although the ACA is already efficient compared with the direct method, this paper utilizes a novel technique to further reduce the setup time and storage memory. This method applies the predetermined interaction list supported oct tree (PILOT) to form a new far field interaction list. Using the new far field interaction list, less setup time representation of the far field matrix is obtained. The numerical results of complex objects are used to demonstrate that the memory requirement of the modified ACA is also less than that of the traditional ACA. An efficient preconditioning technique is combined into the inner-outer flexible generalized minimal residual (FGMRES) solver to further speed up the matrix-vector multiplication. © 2011 ACES.