[Dune] [#928] breakdown of BiCGSTAB

[Markus Blatt]

On Thu, Jun 09, 2011 at 02:58:10PM +0200, Andreas Lauser wrote:
> Hi All,
> 
> I might be wrong (In this case I appologize for the noise), but I always 
> thought that BiCG(Stab) reduces to CG in the symmetric positive definite case. 
> If this is true, then the breakdown of BiCGStab but not CG indicates a problem 
> with the ISTL implementation of BiCGStab, right?
>

This would be the BiCG method. i.e. without the stabilisation. 
BiCGStab can be interpreted as a product of BiCG and GMRES. The latter
causes the problem here because it stagnates and causes a breakdown.

Anyway, the matrix considered is NOT symmetric due to the Dirichlet
boundary considitions represented as DoFs.

Markus
-- 
Markus Blatt, Interdisciplinary Center for Scientific Computing,
University Heidelberg, Im Neuenheimer Feld 368, D-69120 Heidelberg

'Mathematics knows no races or geographic boundaries; for
mathematics, the cultural world is one country' - David Hilbert