StableEvaluationAtAPoint

[erickschulz]

I could modify the code so that now it takes a finite element function. I create it by solving a Dirichlet problem where the boundary values are given by the exact solution. I could also fix the core dump run time error. However, the method for stable evaluation at a point is still not converging. There is still some fixing to do.

Note that the current up to date version of the code is in the associated BRANCH, and not in the main master branch.

Let me know if you catch the problem. I will also go through the details again.

EDIT : I tried with a different set of meshes to make sure they weren't broken like we already experienced, but also obtained bad results.

Erick

[hiptmair]

The description of the problems encountered is too vague to be useful.

[hiptmair]

Suggestions for a less complex test. Clone you implementation of Jstar, replace the MeshFunctionFE with a MeshFunctionGlobal wrapping a harmonic function. This allows to test (3.11.14) directly.

[peterphilippe]

This should now be fixed in pr #31.