Volume 58, pp. 244-270, 2023.

On Cvetković-Kostić-Varga type matrices

Lei Gao and Chaoqian Li

Abstract

Cvetković-Kostić-Varga (CKV)-type matrices play a significant role in numerical linear algebra. However, verifying whether a given matrix is a CKV-type matrix is complicated because it involves choosing a suitable subset of $\{1,2,\ldots,n\}$. In this paper, we give some easily computable and verifiable equivalent conditions for a CKV-type matrix, and based on these conditions, two direct algorithms with less computational cost for identifying CKV-type matrices are put forward. Moreover, by considering the matrix sparsity pattern, two classes of matrices called $S$-Sparse Ostrowski-Brauer type-I and type-II matrices are proposed and then proved to be subclasses of CKV-type matrices. The relationships with other subclasses of $H$-matrices are also discussed. Besides, a new eigenvalue localization set involving the sparsity pattern for matrices is presented, which requires less computational cost than that provided by Cvetković et al. [Linear Algebra Appl., 608 (2021), pp.158–184].

Full Text (PDF) [817 KB], BibTeX

Key words

CKV-type matrices, $S$-Sparse Ostrowski-Brauer type-I matrices, $S$-Sparse Ostrowski-Brauer type-II matrices, $H$-matrices

AMS subject classifications

15A18, 15A42, 15A69

Links to the cited ETNA articles

[9]	Vol. 18 (2004), pp. 73-80 Ljiljana Cvetkovic, Vladimir Kostic, and Richard S. Varga: A new Geršhgorin-type eigenvalue inclusion set

< Back