Major Discussion Issues
List of Problems that Need Discussion
MeshData or EquationSystems can both be used (only one tested) but which strategy provides the best approach.
MeshData
Pros:
- More Natural since we are simplifying defining data at grid locations
- Simpler less heavyweight implementation
Cons:
- No interpolation
- Few people use it so it may be buggy.
- Only seems to work with UNV mesh format (not with xta for sure)
Equation Systems
Pros:
- Works with a variety of mesh types
- Provides a natural means of interpolation
- Used more frequently by libMesh developers so perhaps less buggy.
CellML implementation completed by Jagir
- What is the status and can we get going soon on this?
Overall problem setup (reading in problem description)
- Are we going to use a perl framework? python? just C++?
- How will we use first bullet with this work? Translating all of old examples to new libMesh examples?
- How will users run examples? Can we start thinking about this? It may help direct development of OpenCMISS.
Monodomain/Bidomain Class
- Does it even make sense to create a new bidomain class or is it more natural to just use EquationSystems?
Parallel Mesh Class
- While libmesh did not distribute the mesh (it was copied to each node) when we were evaluating it, this is currently being actively worked on by libmesh developers.