Minutes 16 October 2006
Present: Travis, Jagir, Karl, Ehsan.
Ehsan is modelling transport of solutes in the eye.
- Includes: fluid flow with osmotic gradients, diffusion, conduction.
- Unknowns: concentrations, fluid velocities, electrical potential.
- Looks like this has been considered from a finite-volume bidomain perspective.
- Could probably be done with finite elements. Is this better?
Boundary conditions
- Travis using normal Dirichlet boundary condtions where a column and row
are replaced with those from the Identity and a contribution is made to
the rhs.
- Looks like DenseMatrix::condense may be useful here.
libMesh solution variables
- 3 possibilities
- local unknowns: System.solution
- local unknowns + "ghost" unknowns: System.current_local_solution
- all unknowns: ??
- When are vectors updated from other vectors.
- current_local_solution is updated automatically from solution in ImplicitSystem::solve.
KDevelop
- Jagir and Ehsan have been looking at using KDevelop for writing code to
use LibMesh.
- But KDevelop seems to use autotools to write a Makefile.
- Can this be turned off?
- Or where can the libMesh library location be specified?
- But KDevelop seems to use autotools to write a Makefile.