Quasidirect decompositions of Hankel and Toeplitz matrices (Q799761): Difference between revisions

An \(n\times n\) matrix \(A=[a_{ij}]\) is called a Hankel matrix if the values of its entries are constant along the diagonals \(i+j=m\) for \(m=2,3,...,2n\). If \(A=B+C\) then A is said to be a quasidirect sum of B and C if for some invertible P and Q and suitable square blocks \(B_ 0\) and \(C_ 0\) we have \(B=P\quad diag(B_ 0,0)Q, C=P\quad diag(0,C_ 0)Q\) and \(A=P\quad diag(B_ 0,C_ 0)Q.\) It is easily shown that A is a quasidirect sum if and only if its rank \(r(A)=r(B)+r(C).\) The Hankel matrix A is called H-indecomposable if it cannot be written as a quasidirect sum of two nonzero Hankel matrices. The main result of the paper is to prove that, over an algebraically closed field, every Hankel matrix is a quasidirect sum of H- indecomposable matrices and this decomposition is unique if and only if the matrix is nonsingular (Theorem 2.23). Furthermore all H- indecomposable matrices are described, and some weaker results are proved in the case the field is not algebraically closed. The author notes that, since there is a simple linear transformation from the space of Hankel matrices to the space of Toeplitz matrices, analogous results for Toeplitz matrices also hold.