[SciPy-user] coo_matrix, lapack, cvxopt

Nathan Bell wnbell@gmail....
Tue Mar 24 11:31:03 CDT 2009


On Tue, Mar 24, 2009 at 9:56 AM, Nils Wagner
<nwagner@iam.uni-stuttgart.de> wrote:
>
> A symmetric positive definite matrix is required in case
> of CG.
>
> http://en.wikipedia.org/wiki/Conjugate_gradient_method
>

You can generally apply it to semi-definite problems too, but it
sometimes fails.

scipy.sparse.minres() is a better choice with comparable costs to cg()


-- 
Nathan Bell wnbell@gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


More information about the SciPy-user mailing list