Skip to main content
4 of 4
Broken link fixed.
Joseph O'Rourke
  • 150.9k
  • 36
  • 358
  • 958

If you will excuse me substituting a polyhedron for the Riemannian manifold (imagine rounding the vertices), this figure shows how the cut locus (red) from source $x$ is a tree. (The green arcs are equidistant from $x$.)
     2x1x1Box
     (Figure from Discrete and Computational Geometry)

Concerning, "What are the edges?": They are geodesics, such that there are two distinct shortest paths from $x$ to every interior point of an edge of the cut locus. Points of the cut locus of degree $k$ have $k$ distinct shortest paths from $x$.

Joseph O'Rourke
  • 150.9k
  • 36
  • 358
  • 958