instabilities in visualization

[grosilho]

The visualization process creates a mesh for every domain and then evaluates the functions on this mesh employing the boundary data and the green representation formula. For every node in the mesh, the function is evaluated by computing integrals on the boundary.
The problem is that some instabilities are produced for the mesh nodes very close to the boundary, i.e. the first "layer" of nodes at a distance approximatively dx from the boundary. Moreover, it looks like this phenomenon is independent of dx, since it happens only for nodes at a distance dx from the boundary, disregarding the size of dx.