Nijso

The edge correction in PhysicalGeometry is correctly modifying the normals, all edges between ipoint and jpoint that lie on the symmetry plane are identified, and the normal of the dual...

> So modifying the gradient + residual is an improvement and it's just the normals that cause issues? Or the change to the residual is also problematic? The modification of...

Below is a new testcase, laminar flow around a cylinder. In the line figure I show the normal velocity on the axis y=0, with the solid red line the results...

Symmetry plane is correct now. Below the figure of the super coarse diverging channel flow. Top figure: corrected symmetry, Bottom figure: original. The symmetry plane is in the middle of...

Here is a picture of the diverging channel flow with the pressure contours. This is not related to the current implementation, but notice that at the corner where the outlet...

Update: Everything seems to work correctly now.

Results for laminar flow around a cylinder. right-top figure shows normal velocity at the symmetry axis, which should be zero. old (develop) can show very high values for normal velocity....

Flow around a sphere, taking a 5 degree slice with 2 symmetry planes. Top:new implementation. bottom: old (develop). This picture shows axial velocity, for both converged to (pressure)R=-10. 

TO DO: add testcase with 2 symmetry planes (laminar flow around sphere, slice)

There are some other find_last_of lines as well, if all filenames are supposed to be given without any file extension, then these should all be removed.