How easy is it to parallelize delaunay triangulation? It seems that you would be constrained to operating on independent sets of triangles when flipping edges to ensure correctness
Why is just flipping sufficient to get a Delaunay mesh? It does not seem intuitively true due to many adversarial cases one can think of
Do we have to flip all the edges? It seems like flipping all edges will result in a mesh that is basically the same but in a different orientation.