Let be the matrix of order , generated by the command . Starting from the unitary vector, execute iterations of the inverse power method applied to . The absolute error associated to the approximation of the eigenvalue that is the closest to (take the value returned by eig as reference), is approximately: