This is a read only copy of the old FEniCS QA forum. Please visit the new QA forum to ask questions

Iterative solver for large-scale mixed poisson problem

+1 vote

I am trying to solve the demo mixed-poisson equation (with RT0 elements) in a large-scale setting (i.e., 2000000 triangles). I will always get a solution with a direct solver, but it will be extremely slow. When i try an iterative solver, it doesn't seem to converge. I have tried krylov solvers like gmres and bicgstab and almost every preconditioner available but nothing seems to work. Is there a certain or well-known solver/preconditioner combo that works best for mixed methods like RT0? Thanks

asked Mar 4, 2015 by jychang48 FEniCS Novice (510 points)

1 Answer

0 votes

Probably a domain decomposition preconditioner will work better.

http://www.ima.umn.edu/~arnold//papers/hdiv.pdf

but this is not directly implemented in dolfin yet.

answered Mar 5, 2015 by chris_richardson FEniCS Expert (31,740 points)

Isn't this related to the technique related to the demo stokes-iterative method? Where we would have to create our own multi-grid preconditioner?

commented Mar 6, 2015 by jychang48 FEniCS Novice (510 points)

Have you solved your problem? I also encountered the same problem

commented May 30, 2017 by kingboxing FEniCS Novice (120 points)
...