An add-on question. Member function *mg::SmootherRelaxation::smooth* needs to call *RelaxationType::step* to actually do iterations, which in my case is *PreconditionJacobi*. And *PreconditionJacobi::step* is an interface calling *Jacobi_step* from its MatrixType inside.
However this is not offered in the matrix-free Base (only find *Base::precondition_Jacobi* multiplying vectors with Jacobi preconditioner) like in the sparse matrix class. Seems I need to implement the Jacobi step on my own, just like the way *PreconditionChebyshev* does? Or is there any warped class to accomplish it? 殷承江 <[email protected]> 于2021年10月21日周四 下午11:02写道： > Hi all, > > As described in the title, I am looking for an appropriate preconditioner > for an asymmetric matrix implemented in matrix-free style. > > I've gone though all the matrix-free tutorials and none of them talks > about the asymmetric case. (Almost all of them use the Chebyshev smoother > with different inner preconditioners) > > However, *Chebyshev smoother* is out of the question due to its > requirement of symmetric positive definite matrix to get the max > eigen-value. *BlockRelaxation preconditioner* is also unlikely to fit, > which requires the inner contribution to cell blocks using discontinuous > finite element. Other kinds of relaxation preconditioner like *SOR, ILU* > all request certain entries of the matrix, which is expensive in > matrix-free framework. > > So I guess the only doable choice would be the *PreconditionJacobi + > GMRES on the inner levels and AMG + GMRES solver on the coarse grid*? > > Since I am still learning this amazing library and not familiar with the > framework. Feel free to correct me if I got anything wrong. > > Regards, > Chengjiang Yin > > -- > The deal.II project is located at http://www.dealii.org/ > For mailing list/forum options, see > https://groups.google.com/d/forum/dealii?hl=en > --- > You received this message because you are subscribed to a topic in the > Google Groups "deal.II User Group" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/dealii/uY10wLpBSa0/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/dealii/0f8197d4-6b88-42d0-a2a3-303e05601a08n%40googlegroups.com > <https://groups.google.com/d/msgid/dealii/0f8197d4-6b88-42d0-a2a3-303e05601a08n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/CAGk2Tys5uMY7GC2C-QJTiKtaqug_H6NKaxi9z%3D8_RA1HUFtZ8Q%40mail.gmail.com.
