I am validating my stability code written in python and I need to prove that my eigenvalues are not Krylov size dependent. I am using scipy.sparse.linalg.eigs() routine and I would like to know if I can show converging of eigenvalues with varying Krylov sub-space size. I played with the tol and maxiter settings, however, for my particular problem the leading eigenvalue was always the same, even with extreme values like tol=1 and maxiter=2. Is there a direct way I can manipulate Krylov in SciPy.sparse.linalg.eigs()?
I tried varying "maxiter" and "tol" for an eigenvalue problem with k=5 and all 5 of the eigenvalues never changed even with extreme values of "maxiter" and "tol"