Suppose $S$ is a non-compact Riemann surface in $\mathbb R^3$ that has no boundary and has genus zero (i.e. its fundamental group is generated only by its ends at infinity). A typical example would be a boundary of a tubular neigborhood of a tree, with all leaves on the sphere at infinity.

Is it true that one of the 3-manifolds into which $S$ divides $\mathbb R^3$ is homeomorphic to $\mathbb R^3$? (The answer is clear for the example given above, but how is it in general?)

  • 1
    $\begingroup$ I think you should only consider connected surfaces that are properly embedded in $\mathbb{R}^3$, because otherwise $\mathbb{R}^3\setminus S$ may be connected, and not homeomorphic to $\mathbb{R}^3$ (e.g. when $S$ is the standard sphere with some disjoint closed disks removed). $\endgroup$ Oct 4, 2013 at 13:28
  • $\begingroup$ Sure, in fact I should have written `closed, non compact, no boundary'. Thanks for pointing this out. $\endgroup$ Oct 4, 2013 at 13:56

1 Answer 1


No, it's not true. Here's a counterexample. Consider three concentric infinite cylinders (properly) embedded in the obvious way in $\mathbb R^3$. Join cylinders 1 and 2 with a knotted tube, and similarly join cylinders 2 and 3 with a knotted tube. The surface is now a connected sum of three annuli, and hence (connected) genus zero. A simple homology calculation shows that neither of the two complementary regions is a 3-ball.

Oops. Since the example only depends on a homology calculation, it doesn't matter whether or not the tubes are knotted -- the knots are an unnecessary distraction. (I got distracted by the standard examples of closed surfaces in $S^3$ which don't bound handle-bodies on either side.)

  • $\begingroup$ The fundamental group of each complementary component will contain a knot group as a summand, so cannot be simply-connected. $\endgroup$
    – Ian Agol
    Oct 4, 2013 at 15:38
  • $\begingroup$ @Ian, yes, that was my original idea, but then I realized that there's a simpler homology argument and that the tubes don't need to be knotted. $\endgroup$ Oct 4, 2013 at 16:10
  • $\begingroup$ That is a very nice example, thanks a lot. How about a follow-up question: What if we additionally assume that S is planar, i.e. homeomorphic to an open subset of a plane? I believe that the answer should be YES in that case... $\endgroup$ Oct 8, 2013 at 9:56
  • $\begingroup$ If I understand you correctly, the surface in my example is planar in your sense. A connected sum of three annuli is homeomorphic to a large disk with five smaller disks removed. $\endgroup$ Oct 10, 2013 at 0:25
  • $\begingroup$ Of course you are right, I feel a bit ashamed not to have realized that. Thanks a lot. $\endgroup$ Oct 10, 2013 at 7:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.