Re: [Getfem-users] Using SuperLU - A question

[Yves Renard]

Le Samedi 21 Octobre 2006 00:47, Bwalya Malama a écrit :
> Hi:
>
> I am trying to solve the following using SuperLU:
>
> Ax = b (1)
>
> where the vector "b" changes as I match thru time but the matrix "A" is
> fixed. Using "routines" from numerical recipies, I can solve (1) with "A"
> factored into "L*U" only once. For different "b"'s I only call the
> backsubstitution routine. My question is, can one do the same thing using
> SuperLU? From what is in the gmm++ manual, it appears one has to decompose
> "A" everytime (1) is solved for different "b"'s. Is that the case? Or is
> it possible to decompose "A" only once, and using "LU" for subsequent
> solutions?

Hi,

Yes, you can decompose the matrix and store the decomposition.

The gmm interface to SuperLU allows to declare an object

SuperLU_factor<double> SLUf;

The decomposition is made by the method

SLUf.build_with(A);

Where A is your sparse matrix. Then, you can solve the triangular systems 
using

SLUf.solve(x, b);

(You can also solve the transposed system using SLUf.solve(x, b, 
getfem::LU_TRANSP) and with complexes the conjugated system with 
SLUf.solve(x, b, getfem::LU_CONJUGATED) ).


By the way, The Gmm interface to SuperLU is not complete at all. A lot of 
SuperLU fonctionalities are not interfaced. You can also take it as a model
to build your own interfaces to the methods you need.
 
Yves.

-------------------------------------------------------------------------
  Yves Renard (address@hidden)       tel : (33) 04.72.43.80.11
  Pole de Mathematiques, INSA de Lyon          fax : (33) 04.72.43.85.29
  Institut Camille Jordan - CNRS UMR 5208
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard
-------------------------------------------------------------------------