Skip to main content
2 events
when toggle format what by license comment
Aug 13, 2010 at 15:07 comment added J. M. isn't a mathematician Well I certainly don't need to get all the eigenvalues (as a QR algorithm implementation is wont to do) just for the sole purpose of estimating the spectral radius! I just need a good estimate computable from the entries of the Hessenberg matrix (I am working with dense matrices here; Arnoldi/Lanczos-type methods as I understand are better suited for the sparse case). With the power method, I had mentioned how troublesome convergence can get for matrices with equimodular dominant eigenvalues. If quick estimates of the largest root can be done for polynomials, why not for the spectral radius?
Aug 13, 2010 at 12:08 history answered Federico Poloni CC BY-SA 2.5