it's not symmetric :) i want to go for the matrix-free approach that Timo suggested, but how about preconditioning? can one build a preconditioner using one of the standard deal.ii classes and the new Matrix wrapper that Timo gave below?
thanks! -- Mihai ________________________________ Von: Jon S. Squire. <[email protected]> An: Timo Heister <[email protected]> CC: mihai alexe <[email protected]>; deal.ii <[email protected]> Gesendet: Donnerstag, den 2. Dezember 2010, 10:01:55 Uhr Betreff: Re: [deal.II] solving A^T x = b If A is symmetric, you do not have a problem. :) Jon [email protected] www.csee.umbc.edu/~squire On Thu, 2 Dec 2010, Timo Heister wrote: > Hello Mihai, I got an idea: you can try to wrap SparseMatrix in a class that applies tvmult instead of vmult. Code (not tested, not sure it even compiles): template<class MAT> class TMatrix { public: TMatrix(const MAT &m) : mat(m) {} template<class VECTOR> void vmult (VECTOR &dst, const VECTOR &src) const { mat.Tvmult(dst,src); } private: const MAT & mat; }; Obviously you don't have access to preconditioners like this... -- Timo Heister http://num.math.uni-goettingen.de/~heister On Wed, Dec 1, 2010 at 3:18 AM, mihai alexe <[email protected]> wrote: > Hello all, > What could I use with deal.ii to solve A^T x = b where I have A as a > SparseMatrix<double>? I am now using Tsolve (with SolverRichardson<>), but > even with preconditioning, it is really slow. I would like to use something > like SolverGMRES<> if possible, but I don't see a way to tell it to use A^T > instead of just A. > Your help is very much appreciated :) > Best, > -- Mihai > > _______________________________________________ > dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii > > _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
_______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
