Smoother solvent excluded surface atom boundaries

[Tom Goddard]

Would like to avoid jagged defects in atom boundaries on solvent excluded surfaces. Currently such defects usually arise near the junction of 3 atom surface patches. A single triangle having only 2 closest atoms for the 3 vertices doesn't give any territory within the triangle to the third atom. An obvious case is when two atom patches are separated by a thin strip belonging to a third atom. If the triangle bridges the third atom territory then the patches for the first two atoms across cut the strip. This is seen in numerous places in the attached image.

A possible solution is after adding vertices to make sharp boundaries, recompute the closest atoms and then recompute patch boundaries. This should work but is complicated by the duplication of vertices on the boundaries. Those have equidistant atoms leading to random atom assignments.

I've attached an image showing the current defects for carpaine. Structures like this one with hydrogens are particularly bad.

[Tom Goddard]

Done some months ago. Looks beautiful. Iteratively subdivides triangles.