FiPy: A Finite Volume PDE Solver Using PythonMULTIGRID and ADAPTIVE MESHING
This page is to figure out what we have to do to fipy to get multigrid and adaptinve meshing working.
Questions that need answering
- Should we pull in an algebraic multigrid solver before
embarking on a geometric multigrid?
- Is it useful to implement geometric multigrid without
having adaptive griding?
- Have we noticed a trend in code use in the phase field
community?
Review of packages
The first step has to be to figure out which code base to start from. Are they Python wrapped? How would they fit in? Would it require updating Variable, Mesh and Terms in a big way? Can the package work mostly inside our paradigm?
PyTrilinos
http://software.sandia.gov/trilinos/packages/pytrilinos/
Trilinos comes with the ML package that Roman wrapped at one point and so is probably worth looking at.
MG net
Paramesh
http://www.physics.drexel.edu/~olson/paramesh-doc/Users_manual/amr.html http://www.physics.drexel.edu/~olson/paramesh-doc/Users_manual/multigrid.html http://sourceforge.net/projects/paramesh/
The multigrid support is still new. Distributed under the NASA license, like NIST probably. Licence is completly open.
Chombo
http://seesar.lbl.gov/ANAG/chombo/
Not Python.
PyTrilinos
http://software.sandia.gov/trilinos/packages/pytrilinos/index.html
Has mutigrid and python wrappers
Pyro
http://www.ucolick.org/~zingale/pyro/
Gerris
Has multigrid.
PETSc
http://www-unix.mcs.anl.gov/petsc/petsc-as/
Don't see any multigrid solvers in Petc.
Other Resources
Need a link here StoreHere?
