[SciPy-user] spsolve

[Nathan Bell]

On Wed, Jul 23, 2008 at 3:46 PM, Christian Bähnisch
<christian-baehnisch at gmx.de> wrote:
>
> I'm using "scipy.linsolve.spsolve". Unfortunately, I have not been able to
> figure out if  it is possible to get the number of iterations actually
> used by the method. I would be very glad if somebody could help me on
> this topic
>

Christian, the sparse solvers in scipy.linsolve are what is known as
"direct" methods (as opposed to "iterative" methods).  Rather than
iteratively solving a linear system (like the Conjugate Gradient
method), direct solvers just compute a sparse LU factorization of the
matrix in order to solve the system.  In order to do this efficiently,
and to keep L and U as sparse as possible, they incorporate a number
of fairly sophisticated algorithms/heuristics.

In SciPy 0.6 scipy.linsolve ultimately calls either the SuperLU or
UMFPACK packages to perform the solve.

If you're interested in iterative solvers, there are several in
scipy.linalg.iterative.

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