Implementing the Stokes equation with PSPG stabilization

variational formulation
5

I still can’t make the solver converges, but I come up with the following hypotheses:

  • I think that to use u instead of w.split()[0], is necessary to change the solver configuration. I have tried to use GMRES with different preconditioners, but I still get the same error. Reading some topics in the group I noticed that a similar problem occurred with the SUPG stabilization link.

  • Is there a command in FEniCS similar to COMSOL’s nojac()?. I’m asking that because I think that use w.split()[0] might have a similar effect as nojac(), since w.split()[0] it is a deepcopy of the w and is probably not contributing to the Jacobian which is calculated with respect to w.

I’m insisting on this topic because I’m going to work with a more complex stabilization parameter that involves other variable and their gradients. That way, I would be very grateful if someone could help me with this topic.