[SciPy-user] Sparse eigenvalues/eigenvectors problem
Thu Jun 12 16:52:52 CDT 2008
On Thu, Jun 12, 2008 at 9:24 AM, James Philbin <firstname.lastname@example.org> wrote:
> So it seems that eigen_symmetric is failing to find lots of the
> repeated 1 eigenvectors. Is this an inherent problem with ARPACK or
> symptomatic of a bug in scipy? Matlab's eigs seems to not suffer from
> this nearly as much, returning nearly all ones for both the 10 and 20
> cases. I assume this is also using ARPACK under the surface. I can
> upload the matrix S somewhere if people are interested.
While your problem could be caused by a bug in our ARPACK wrappers, it
could also be due to differences in default parameters. Are you sure
that MATLAB is calling ARPACK with the same parameters as
eigen_symmetric? For instance, it the same tolerance used in both
cases? Have you setup eigs() to use the symmetric algorithm?
There may also be a difference in the way MATLAB and SciPy chose the
initial vector of the Krylov subspace (parameter v0 in eigen()).
If you believe this to be a genuine bug, please submit a ticket in
Trac with a short script that demonstrates the error.
If your matrix is small enough, you could include it with the ticket
as well, perhaps as a compressed MatrixMarket file:
>>> A = #your matrix
>>> from scipy.io import mmwrite
>>> mmwrite('A.mtx', A)
$ gzip A.mtx
> I also think i've fixed a bug in the current version of arpack.py
> relating to the info returned (The warning that maxiters had been
> reached was never shown) and added a warning if some eigenvectors
> didn't converge:
Added in r4441:
Nathan Bell email@example.com
More information about the SciPy-user