[SciPy-User] Eigenvectors of sparse symmetric matrix
Mon Oct 25 13:23:57 CDT 2010
On Mon, Oct 25, 2010 at 2:14 PM, Hao Xiong <firstname.lastname@example.org> wrote:
>>> I am trying to compute the eigenvectors corresponding to the d+1
>>> smallest eigenvalues of A=W.T*W. I started with W as a dense matrix and
>>> W = sparse.csr_matrix(W)
>>> A = W.dot(W) # W.T * W
>> That is W*W and not (W.T)*W
> Thanks, Pauli. Somehow I convinced myself it was otherwise. I have corrected that.
>>> W,V = eigen_symmetric(A,d+1, which='SM')
>>> The biggest problem is that the algorithm fails to converge and I get
>>> all zeros as eigenvectors for a testing dataset. Using dense SVD I got
>>> the expected results.
>> You can try playing with setting the maxiter parameter to allow ARPACK to
>> spend more iterations on the problem.
> I tried maxiter=100000 and still got zero vectors. I must be missing something.
just a weird idea, since I have no idea what eigen_symmetric is doing,
and there are no docs that I have seen for the extra options:
Is it possible to run a dense svd on a (random) subset of the data and
then use those as starting values for the sparse decompositions?
> SciPy-User mailing list
More information about the SciPy-User