当前位置: X-MOL 学术Linear Multilinear Algebra › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
New upper bounds for the spectral variation of a general matrix
Linear and Multilinear Algebra ( IF 1.1 ) Pub Date : 2020-09-20 , DOI: 10.1080/03081087.2020.1822274
Xuefeng Xu 1
Affiliation  

Let ACn×n be a normal matrix with spectrum {λi}i=1n, and let A~=A+ECn×n be a perturbed matrix with spectrum {λ~i}i=1n. If A~ is still normal, the celebrated Hoffman–Wielandt theorem states that there exists a permutation π of {1,,n} such that (i=1n|λ~π(i)λi|2)1/2EF, where F denotes the Frobenius norm of a matrix. This theorem reveals the strong stability of the spectrum of a normal matrix. However, if A or A~ is non-normal, the Hoffman–Wielandt theorem does not hold in general. In this paper, we present new upper bounds for (i=1n|λ~π(i)λi|2)1/2, provided that both A and A~ are general matrices. Some of our estimates improve or generalize the existing ones.



中文翻译:

一般矩阵谱变化的新上界

一个Cn×n是具有谱的正规矩阵{λ一世}一世=1n, 然后让一个~=一个+Cn×n是一个有谱的扰动矩阵{λ~一世}一世=1n. 如果一个~仍然是正常的,著名的Hoffman-Wielandt 定理指出存在一个置换π{1,,n}这样(一世=1n|λ~π(一世)-λ一世|2)1/2F, 在哪里F表示矩阵的 Frobenius 范数。该定理揭示了正规矩阵谱的强稳定性。但是,如果A一个~是非正态的,Hoffman-Wielandt 定理一般不成立。在本文中,我们提出了新的上限(一世=1n|λ~π(一世)-λ一世|2)1/2,前提是A一个~是一般矩阵。我们的一些估计改进或概括了现有的估计。

更新日期:2020-09-20
down
wechat
bug