Minutes 3 July 2006

Present: Shane, Karl, Travis, Mark.

MPI Test problem proposal and implementation

Should we fix up cm?
- There are at least 4 areas that could be fixed up
  - lines
  - faces
  - sparsity calculation
  - something in the solve that is not the linear solve and not the assembly.
- No one present thinks fixing cm should be a priority, but Chris seemed to think it should be fixed.
- If someone wants to fix up cm then there is no harm in doing so.
Should we set up a comparable grid-based fem problem?
- Grid-based fem not as flexible as what libmesh provides.
- Grid-based fem is what we would use for a reaction diffusion system.
- Not sure how much work in changing a cell model to one with infinite resistance to turn a reaction diffusion system into a diffusion only system.
- But can set up a different but similar problem to give an indication of time required to solve: monodomain problem with the simplest available cell model and simple Euler integration.

Libmesh and visualization

Selecting a package

Could spend up to a week looking into each of the other shortlisted packages.
Things we want to check with libmesh developers:
- Cubic Hermite algorithms.
- Embedded meshes.
- Shane: non-linear mechanics.
- Karl: possibly fibre angles for anisotropic solutions.
If libmesh is selected, need to set up a plan of how to build on it.

SUR cluster

Resourcing

MPI job candidates
- next stage seems to be getting more information and references.
cmiss.org development
- Carey's replacement plan had been canned then reinstated but replaced with a physiome.org replacement that might not be interested in cmiss.