SINGULAR VALUE STRUCTURE OF REAL MATRICES WHICH CAN BE EXPRESSED AS A LINEAR COMBINATION OF TWO ORTHOGONAL MATRICES

Zheng Li,Tie Zhang,Chang-Jun Li
DOI: https://doi.org/10.7153/oam-2022-16-55
2022-01-01
Operators and Matrices
Abstract:Can every real matrix be expressed as a linear combination of a certain number of orthogonal matrices? What is the smallest number of these orthogonal matrices? These elegant and interesting questions were initially raised by Zhan [7]. Soon afterwards, Li and Poon [4] proved that kmin , the smallest number of these orthogonal matrices, is not greater than 4. These classic results inspire us to further explore the improvement or supplement of this theory. We investigate some fundamental properties of An(k), which is the set of all n x n real matrices that can be expressed as a linear combination of k orthogonal matrices. Furthermore, we characterize the singular value structure of matrices in set An(2) and the block structure of related orthogonal matrices. We obtain an equivalent condition and some sufficient or necessary conditions of A is an element of An(2). Based on these results, we demonstrate the existence of matrices that are not in set An(2), and prove that kmin > 2 (for n 3).
What problem does this paper attempt to address?