I am trying to solve an incompressible navier stokes on an adaptive mesh,
and I find that using a petsc fieldsplit preconditioner which works
reasonably well on a normal mesh performs very very poorly on an adaptive
mesh. My solver options are as follows
Is there anything in particular about the matrices generated in the navier
stokes problem that leads it to fail badly.. ( I usually see the solver
converge in 2 steps but with the adaptive mesh, it converges linearly) .
I have an associated cahn hilliard solver, and that has no problem when
solved after adaptivity.
Subramanya G Sadasiva,
Graduate Research Assistant,
Hierarchical Design and Characterization Laboratory,
School of Mechanical Engineering,
"The art of structure is where to put the holes"
Robert Le Ricolais, 1894-1977
From: Roy Stogner <roystgnr@ic...> - 2013-04-04 21:45:15
On Thu, 4 Apr 2013, Subramanya Sadasiva wrote:
> Is there anything in particular about the matrices generated in the navier
> stokes problem that leads it to fail badly.. ( I usually see the solver
> converge in 2 steps but with the adaptive mesh, it converges linearly) .
Are you using the symmetric or asymmetric constraints option? You
might try the alternative.
What are your FE variable types?