首页> 外文会议>Progress in electromagnetics research symposium (PIERS 2009 Moscow) >On the Relationship between Nonuniqueness of Electromagnetic Scattering Integral Equations and Krylov Subspace Methods
【24h】

On the Relationship between Nonuniqueness of Electromagnetic Scattering Integral Equations and Krylov Subspace Methods

机译:电磁散射积分方程的非唯一性与Krylov子空间方法之间的关系

获取原文
获取原文并翻译 | 示例

摘要

Some relevant operator equations used to describe natural phenomena can degenerate. This is actually the case of the electric (EFIE) and magnetic (MFIE) field integral equations in scattering theory, both failing to provide a unique solution (at least in a standard sense) in correspondence of some set of frequencies (known as resonances), for which they lack injectivity.rnOne of the most common expedient to the problem deals with the concept of a generalized solution (say a pseudo-solution), whose computation is numerically related to the Moore-Penrose pseudo-inverse of some large and generally dense square matrix A ∈ C~(n,n) resulting by the discretization of the integral model via the Method of Moments (MoM) or different projective schemes and its consequent approximation (in the sense of the uniform operator limit) by means of a linear system of algebraic equations such as Ax = b. Ultimately this involves the very time-consuming task of computing the singular value decomposition (SVD) of A, in order to filter the noisy effects introduced on the unknown current by the almost-singularity of A when EFIE or MFIE are applied in the neighborhood of a resonance.rnIn this paper, we try to give an answer and provide theoretical motivations and numerical evidence to prove it should be the right one, relying on the inherent capability of some Krylov subspace methods to extract the minimum norm solution to the linear system Ax = b to the extent that b lies in the range of A.
机译:用于描述自然现象的一些相关算子方程可能会退化。散射理论中的电场(EFIE)和磁场(MFIE)场积分方程实际上就是这种情况,两者均未能提供与某些频率集相对应的唯一解决方案(至少在标准意义上)(称为共振)这个问题最普遍的解决方法之一是广义解(例如伪解)的概念,广义解的计算与某些大且普遍的Moore-Penrose伪逆在数值上相关。密度平方矩阵A∈C〜(n,n)是通过矩量法(MoM)或不同的投影方案对积分模型进行离散化而得到的,并且由此可以通过a逼近(就统一算子极限而言)代数方程的线性系统,例如Ax = b。最终,这涉及到计算A的奇异值分解(SVD)的一项非常耗时的任务,以便滤除在将EFIE或MFIE应用到A的邻域时,由于A的几乎奇异性对未知电流引入的噪声影响。本文试图给出一个答案,并提供理论依据和数值证据来证明它是正确的,它依靠某些Krylov子空间方法的固有能力来提取线性系统Ax的最小范数解= b在b处于A范围内的范围内。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号