On functions of companion matrices (Q1094500)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On functions of companion matrices |
scientific article |
Statements
On functions of companion matrices (English)
0 references
1987
0 references
The author deals with various representations related to the matrix function f(C) of a complex companion matrix C of \(p(\lambda)=(\lambda - \lambda_ 1)...(\lambda -\lambda_ n).\) These representations involve divided differences of the function \(f(\lambda)\), the power functions as well as the adjoint polynomials associated with C, evaluated at the eigenvalues \(\lambda_ 1,...,\lambda_ n\) of C. Thus he derives such a representation for entries of f(C) (Th. 2.1). Then a representation for f(C) and the similarity-transformation matrix (and its inverse) when C is similar to a block diagonal matrix, each block being associated with a cluster of eigenvalues of C, where different clusters have no eigenvalue in common (Th. 3.1). The special case when the eigenvalues within clusters coincide yields the Jordan representation of f(C) (Cor. 3.1). Another particular case when we have one eigenvalue cluster (Cor. 3.2) gives for \(f(\lambda)\equiv \lambda\) the representation of the similarity-transformation matrix between C and the bidiagonal matrix \((\lambda_ i\delta_{ij}+\delta_{i,j-1})_{n\times n}.\) Combining Th. 3.1. and Cor. 3.2. the author obtains a representation of the similarity-transformation matrix reducing C to a direct sum of lower- order companion matrices (Cor. 3.3). Finally the author gives some criteria for the similarity of f(C) and \(C_{(f)}\), the companion matrix of \(p_ f(\lambda)=(\lambda -f(\lambda_ 1))...(\lambda -f(\lambda_ n))\) (Th. 4.1) and the representation of all similarity-transformation matrices that reduce \(f(C)\) to \(C_{(f)}\) (Th. 4.2 and Cor. 4.1).
0 references
matrix function
0 references
complex companion matrix
0 references
divided differences
0 references
similarity-transformation matrix
0 references
Jordan representation
0 references