Computing Moore-Penrose inverses of Toeplitz matrices by Newton's iteration (Q1764965)

The aim of this paper is to present an algorithm to compute the Moore-Penrose inverse of a rank-deficient Toeplitz matrix. The algorithm is based on Newton's iteration and the concept of \(\varepsilon\)-displacement rank. A simple residual error bound is derived. The bound contains the errors due to Newton's iteration and the errors in the approximation of the displacement representation of the Moore-Penrose inverse. Finally, numerical experiments involving matrices of order up to 32768 are reported, illustrating the convergence of the algorithm.