【摘要】 为了减少预修正快速傅立叶变换算法(P-FFT)或拟合格林函数快速傅立叶变换算法(FG-FFT)的稀疏系数矩阵所需的存储空间,通过改进系数方程的求解方法,获得实系数解.并将改进的求解方法与P-FFT和FG-FF相结合用于计算电场积分方程.所提方案将P-FFT/FG-FFT的稀疏系数矩阵的存储量降到自适应积分方法(AIM)/积分方程快速傅立叶变换算法(IE-FFT)相同水平的同时,未增加矩阵向量积所需FFT的次数,并保持原有算法的精度水平.此外,在每次迭代的时间耗费方面,新方案与AIM/IE-FFT相当.数值实验证实了新方案的上述优点.更多还原

【Abstract】 In order to reduce the storage amount for the sparse coefficient matrix in pre-corrected fast Fourier transform( P-FFT) or fitting the Green function fast Fourier transform( FG-FFT), the real coefficients are solved by improving the solution method of the coefficient equations. The novel method in both P-FFT and FG-FFT for the electric field integral equation( EFIE) is employed. With the proposed method, the storage amount for the sparse coefficient matrix can be reduced to the same level as that in the adaptive integral method( AIM) or the integral equation fast Fourier transform( IE-FFT). Meanwhile, the new algorithms do not increase the number of the FFTs used in a matrix-vector product, and maintain almost the same level of accuracy as the original versions. Besides, in respect of the time cost in each iteration,the new algorithms have also the same level as AIM( or IEFFT). The numerical examples demonstrate the advantages of the proposed method.更多还原