BACKWARD ERRORS AND PSEUDOSPECTRA FOR STRUCTURED NONLINEAR EIGENVALUE PROBLEMS

被引：6
|
作者：
Ahmad, Sk. Safique [1 ]
Mehrmann, Volker [2 ]
机构：
[1] Indian Inst Technol Indore, Sch Basic Sci, Discipline Math, Indore 452020, Madhya Pradesh, India
[2] TU Berlin, Inst Math, Ma 4-5,Str 17,Juni 136, D-10623 Berlin, Germany
来源：
OPERATORS AND MATRICES | 2016年 / 10卷 / 03期
关键词：
nonlinear eigenvalue problem; backward error; symmetric/skew symmetric eigenvalue problem; Hermitian/skew-Hermitian eigenvalue problem; MULTIPLE-EIGENVALUES; CRITICAL-POINTS; MATRIX; LINEARIZATIONS; STABILITY;
D O I：
10.7153/oam-10-31
中图分类号：
O1 [数学];
学科分类号：
0701 ; 070101 ;
摘要：
Minimal norm structured perturbations (backward errors) are constructed such that an approximate eigenpair of a nonlinear eigenvalue problem is an exact eigenpair of an appropriately perturbed problem. Structured and unstructured backward errors are compared. These results extend previous results for (structured) matrix polynomials to more general functions.
引用
收藏
页码：539 / 556
页数：18
相关论文
共 50 条
12345