christopherbatty
commented
11 years ago Actually I seem to be wrong. The sum of tangential and initial normal displacement is used in choosing whether to flip the contract/expanding flag, which affects the resulting normal_displacement. That's pretty subtle; is it intended?
Inside of the face offsetting code, null space smoothing is called to get a tangential offset, but then that seems to be completely overwritten by the normal offset computing using the entropy solution. That should either be fixed, or the null space smoothing pass stripped out of it.