A comparison of the CG-FFT method and the recursive aggregate T-matrix algorithm (EM wave scattering)

A comparison of the computing performances of the conjugate-gradient fast Fourier transform (CG-FFT) method and the recursive aggregate T-matrix algorithm (RATMA) is presented. The advantages of each method are discussed. It is shown that, when the biconjugate CG-FET (BiCG-FFT) works, even for lossless cases, its CPU time performance is quite comparable with RATMA. Since the FFT is a highly vectorizable scheme, implementation of CG-FET or BiCG-FFT on a parallel processing machine will increase the efficiency and save much computing time. However, that this is only for a single incidence angle. If different incident fields are considered, one needs to go through the whole process again for CGM or BiCGM. However, RATMA is valid for any incident field from any angle.<>