[aspect-devel] backward advection

[Wolfgang Bangerth]

On 01/18/2016 10:54 AM, Timo Heister wrote:
> The problem you are running into is that a negative delta_t will make
> the linear systems negative definite. This breaks linear solvers like
> CG. It might be easier to work with a positive delta_t than trying to
> fix all places that will be messed up because of this.

But we don't use CG any more for the advection solver (we use GMRES instead). 
I think what you need to do is not only switch off the physical advection, but 
also flip the sign of the artificial viscosity.

Best
  W.

-- 
------------------------------------------------------------------------
Wolfgang Bangerth               email:            bangerth at math.tamu.edu
                                 www: http://www.math.tamu.edu/~bangerth/