THE POLYNOMIAL EIGENVALUE PROBLEM IS WELL CONDITIONED FOR RANDOM INPUTS

The condition number of the polynomial eigenvalue problem is a measure of the sensitivity of eigenvalues when the input matrix polynomial is slightly perturbed. In this paper we show that the expected value of this condition number for random complex Gaussian inputs can be computed exactly, and actually yields a closed, neat formula!

Based on a paper by Diego Armentano and Carlos Beltrán.

Published in SIAM Journal on Matrix Analysis and Applications.

DOI: 10.1137/17M1139941