Re: [SciPy-User] scipy.sparse.linalg.eigs is faster with k=8 than with k=1