Algorithms for the matrix \(p\)th root (Q2501233): Difference between revisions

Let \(A\) be a real or complex matrix of order \(n\) with no eigenvalues on the closed negative real axis, and let \(p\) be a positive integer. The matrix \(X=A^{1/p}\) is called the principal matrix \(p\)th root of \(A\). Here the authors present new theoretical results and new algorithms for the computation of matrix \(p\)th root. \ At present, a MATLAB implementation of the Schur method, suggested by M. I. Smith, is available as function \textit{rootm }in the Matrix Computation Toolbox by N. J. Higham. Numerical comparisons of the proposed new algorithms against this Schur method are presented and show that sometimes numerical instability affects these new algorithms. The conclusion is that more investigation is needed to improve their finite precision implementation.