Closed maximilianmetti closed 8 years ago
ping @arthbous
This seems to work. I think that there aren't enough degrees of freedom allocated at the cusps on the boundary, which is causing a problem in getting an accurate solution.
Cusps make the H1-norm of the solution pretty large (requiring a lot of elements there to resolve the high frequencies in the gradient). Furthermore, large values on the boundary condition further increase the H1-norm... It seems like the mesh adaptivity is not selective enough where it puts the degrees of freedom.
This PR is able to solve the battery problem with Dirichlet BCs. Still have to figure out the periodic stuff...