Application of the Stochastic Second-degree Iterative Method to EM Scattering from Randomly Rough Surfaces
For a deterministic linear system, if the spectrum of the first degree iterative process lies within the circle centered at one half with radius one half in the complex plane, then the simple stationary second degree method due to Pillis can improve the asymptotic convergence rate over the first degree method. For linear systems with random matrices, however, Pillis approach has to be modified. One recent modification is the Stochastic Second Degree (SSD) method. This paper presents an application of the SSD method to electromagnetic scattering from randomly rough surfaces. When in combination with the popular banded matrix canonical grid (BMIA/CAG) method for two-dimensional scattering for PEC case, with the Jacobi-Richardson shift preconditioning, the resulting SSD-BMIA method can improve convergence for the outer iteration over while maintaining identical convergence properties for the inner iteration of the BMIA method. The computational attractiveness of the BMIA method is well preserved. The cost for the convergence improvement is only an additional storage of 3 n-vectors, or in order in general, where is the number of total unknowns. Numerical illustrations are presented to illustrate the e?ectiveness of the proposed SSD-BMIA method.
Y. Du J. A. Kong
The Electromagnetics Academy at Zhejiang University, Zhejiang University, Hangzhou 310058, China Massachusetts Institute of Technology, Cambridge, MA 02139, USA
国际会议
Progress in Electromagnetics Research Symposium 2007(2007年电磁学研究新进展学术研讨会)(PIERS 2007)
北京
英文
2082-2085
2007-03-26（万方平台首次上网日期，不代表论文的发表时间）