An OpenMP Parallelized Multilevel Green's Function Interpolation Method Accelerated by Fast Fourier Transform Technique

Abstract

A parallelized multilevel Green's function interpolation method (MLGFIM) accelerated by fast Fourier transform (FFT) technique is proposed. The difficulties in applying various improved interpolation schemes to effectively reduce the number of interpolation points are overcome by using the FFT technique. In order to accelerate the convergence property of the iterative solution using the proposed algorithm, a recently proposed preconditioning scheme, i.e., multiplicative Calderon preconditioner is adopted to transform the first kind integral operator to the second kind, albeit an increase of computer memory storage requirement. An OpenMP parallel implementation of the MLGFIM-FFT algorithm on a share-memory computer system is developed to analyze various electrically large electromagnetic scattering problems including a NASA almond, a 20-wavelength cylinder capped with two half spheres, and a 37-wavelength cylinder array. Numerical results illustrate good computational performance of the proposed algorithm.