The usual embedding of the closed orientable surface $M_g$ of genus $g$ in $\mathbb{R}^3$ bounds a compact orientable $3$-manifold which is homotopically equivalent to a bouquet of $g$ circles.

If a compact orientable $3$-manifold $N$ has $M_g$ as its boundary, is it true that $N$ is homotopically equivalent to a bouquet of $g$ circles?


No. As an obnoxious counterexample, take $g=0$ and $N$ to be the complement of some small open ball in a 3-manifold $M$; then $\pi_1(N) = \pi_1(M)$, so if $M$ was not simply connected (whence $S^3$, thanks to Perelman), then $N$ is not homotopy equivalent to the point. More generally, delete (a small neighborhood of) a wedge of $g$ circles from some 3-manifold $M$; the result has boundary $\Sigma_g$, but if $H_1(M)$ is large enough, you couldn't possibly have $H_1(N) = \Bbb Z^g$.

Actually, if $N$ is homotopy equivalent to a wedge of $g$ circles, then it's homeomorphic to the handlebody with $g$ handles, $H_g$; pre-Perelman, one could prove that it's homeomorphic to $H_g \# Y$, $Y$ some simply connected closed 3-manifold.

  • $\begingroup$ do you think there exists any compact manifold with such properties(what OP initially asked)? $\endgroup$ – Anubhav Mukherjee May 26 '16 at 15:22
  • $\begingroup$ @Anubhav.K With what properties? $\endgroup$ – user98602 May 26 '16 at 15:23
  • $\begingroup$ I still do not understand your question. Can you spell it out for me? $\endgroup$ – user98602 May 26 '16 at 15:28
  • $\begingroup$ actually leave it...what initially I was thinking turns out to be wrong... $\endgroup$ – Anubhav Mukherjee May 26 '16 at 15:30

Definitely not. For instance, you could take your example $N\subset\mathbb{R}^3$ and form a connected sum $N\mathbin{\#} P$ where $P$ is any closed oriented 3-manifold. By some simple long exact sequence computations you can find that $H_2(N\mathbin{\#}P)\cong H_2(P)$, so $N\mathbin{\#}P$ cannot be homotopy equivalent to a wedge of circles if $H_2(P)$ is nontrivial.

  • $\begingroup$ I think you mean to have $H_2(P) \oplus H_2(N)$ on the right side of that isomorphism. $\endgroup$ – user98602 May 25 '16 at 3:52
  • $\begingroup$ In general, yes, but in this case $H_2(N)=0$ since I am taking $N$ to be the example OP mentioned. $\endgroup$ – Eric Wofsey May 25 '16 at 3:54
  • $\begingroup$ I misread. Thanks for clarifying. $\endgroup$ – user98602 May 25 '16 at 3:55

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.