[bentivegna]

Hi!

I would like to ask a question regarding the solution of generic non-linear elliptic problems of the type:

laplacian(phi) - C*phi^n = 0

in a cubic box with cyclic boundary conditions in OpenFOAM. I've experimented with a few solver/preconditioner combinations, but everything except the smoothSolver/GaussSeidel option leads to a runaway behavior.

I've read here:

http://www.cfd-online.com/Forums/ope...-openfoam.html

that the linearization of the discretized system is carried out in a "decoupled" way, i.e. performing a Newton's step in each variable separately, while keeping all the other variables fixed and equal to their old value. Does anyone know whether this method is compatible with cyclic boundaries?

(Notice that relaxation methods would be unaffected by this problem, as they do not involve a linearization step.)

Any comment would be very appreciated!