[parallel] Convergence problems for parallel linear solvers

Bug report

The linear solver (e.g. BiCGSTAB with SSOR) has problems when cell-centered methods are applied for non-linear systems on complex domains.

What happened / Problem description

The linear solver has problems when solving the shallow water equations on grids with a complex geometry (non-rectangular domains). The problem often only converges if one iteration for the preconditioner is applied.

What you expected to happen:

The linear solver should also converge if no/more than one iteration is applied.

How to reproduce it (as minimally and precisely as possible):

I will create an example for the shallow water equations to discuss the problem.

Anything else we need to know?:

I couldn't locate the problem, maybe it would be also good idea test the linear solver for a simple linear equation.

Environment:

• Dune version: master and dune-istl feature/parallel-solver-factory
• DuMux version: feature/richards-istl-factory
• Others: