4
$\begingroup$

Stallings' celebrated Fibration Theorem states that if a closed irreducible $3$-manifold $M$ admits a short exact sequence \begin{equation} 1 \to N \to \pi_1(M) \to \mathbb{Z} \to 1, \end{equation} where $N$ is finitely generated, then $M$ fibers over $S^1$.

My question is that whether it is possible to give the exact form of $M$ if we already know that $N$ is a surface group. More precisely, assume there is a short exact sequence \begin{equation} 1 \to \pi_1(S) \to \pi_1(M) \to \mathbb{Z} \to 1, \end{equation} where $S$ is a closed surface. Let $t \in \pi_1(M)$ be a preimage of $1 \in \mathbb{Z}$. Conjugation with $t$ induces an automorphism $\varphi$ of $\pi_1(S)$ whose projection $[\varphi] \in Out(\pi_1(S))$ in the outer automorphism group does not depend on the choice of $t$. By the Dehn-Nielsen-Baer Theorem the extended mapping class group $MCG^{\pm}(S)$ is isomorphic to $Out(\pi_1(S))$. Let $f:S \to S$ be a homeomorphism so that $[f] \in MCG^{\pm}(S)$ corresponds to $[\varphi] \in Out(\pi_1(S))$ under the Dehn-Nielsen-Baer isomorphism. My question is whether $M$ is diffeomorphic to the mapping torus $T_f$ of $f:S \to S$. If this is true, does it just follow from Stallings' proof? Or, in case it is known but requires more work, is there a good reference for it?

More concretely, I am interested in the following situation (which might simplify things). If $M$ is a closed hyperbolic $3$-manifold so that $\pi_1(M)$ is (abstractly) isomorphic to the fundamental group $\pi_1(T_f)$ of a mapping torus, then $M$ is diffeomorphic to $T_f$. This is, of course, an easy consequence of the Geometrization Conjecture. However, I would like to know whether there is a proof of this fact that does not depend on the solution of the Geometrization Conjecture nor on Thurston's mapping torus theorem.

$\endgroup$
3
  • 2
    $\begingroup$ You certainly don't need to invoke Stallings here, since you already assumed the kernel is a surface group. Neither do you need to invoke the full strength of Perelman's geometrisation: Thurston proved that any mapping torus is geometrisable, and in particular in your case $T_f$ has to be hyperbolic. Mostow rigidity then implies that $M$ and $T_f$ are isometric. Alternatively, I think you can avoid geometry entirely using Waldhausen's theorem (see here: mathoverflow.net/questions/242276/…). $\endgroup$
    – HJRW
    Oct 31, 2022 at 15:15
  • $\begingroup$ I explicitely stated that I do not want to make use of Thurston's mapping torus result. But I will have a look at Waldhausen's theorem. Also, I would still be interested in the answer to my first question. $\endgroup$ Oct 31, 2022 at 16:14
  • $\begingroup$ Apologies, I missed that you wanted to avoid Thurston. In that case, I think Waldhausen is probably what you want. If your "first question" is the case of a general surface-by-cyclic group, as opposed to the hyperbolic case, then I think this should also follow from Waldhausen. $\endgroup$
    – HJRW
    Oct 31, 2022 at 16:27

1 Answer 1

3
$\begingroup$

Your question ("is $M$ homeomorphic to $T_f$?") is answered in the affirmative by Theorem 2 of Stallings' paper *On fibering certain 3-manifolds". You will also need his Theorem 1. Here are the statements (slightly simplified).

Theorem 1: Suppose that $M$ is a compact connected three-manifold. Suppose that $\Gamma$ is a finitely generated normal subgroup of $\pi_1(M)$ whose quotient group is $\mathbb{Z}$. Then there is a surface $F$ properly embedded in $M$ so that $\Gamma = \pi_1(F)$.

Theorem 2: With hypotheses as in Theorem 1. Suppose that $M$ is irreducible. Suppose that $\Gamma$ is not $\mathbb{Z}/2\mathbb{Z}$. Then $M$ is a surface bundle over the circle, with $F$ isotopic to a fibre.

That is, your hypothesis on the short exact sequence (plus Theorem 1) gives the surface $F$. Your hypothesis that the manifold $M$ is hyperbolic then gives the additional hypotheses of Theorem 2.

Note that Stallings does not cite Waldhausen. I suppose that this is because his situation is a very very simple case of a Haken hierarchy. Once you have $F$ in your hands (and all the group theory hypotheses), it is "easy" to show that $M - F$ is homeomorphic to a product.

$\endgroup$
10
  • $\begingroup$ Thank you for your answer. I'm not sure if I am obsessing over tiny details, but I have the impression that what is stated is not yet enough to conclude what I want. Theorem 2 gives me that $M$ is some mapping torus $T_{f^\prime}$. However, I would like to conclude that $M$ is $T_f$ with the exact $f$ I described in my question. But I guess this should be clear if one of the following two statements is true (which I state in the next comment): $\endgroup$ Oct 31, 2022 at 19:31
  • $\begingroup$ 1) Let $p:M \to S^1$ be the bundle projection given by Theorem 2. Then the induced map $\pi_1(p):\pi_1(M) \to \pi_1(S^1)$ coincides (up to possibly precomposing with an automorphism of $\pi_1(M)$) with the given surjection $\pi_1(M) \to \mathbb{Z}$. I hope that this is an artefact of Stallings' proof. Is that correct? $\endgroup$ Oct 31, 2022 at 19:38
  • 2
    $\begingroup$ Stallings proves (1) and Waldhausen proves (2). $\endgroup$
    – Sam Nead
    Oct 31, 2022 at 19:53
  • 2
    $\begingroup$ When I say "Stallings proves" I mean "You can find this in the second sentence of section two of his paper (on page 93)" A copy of the paper is available here: maths.ed.ac.uk/~v1ranick/papers/stallfib.pdf $\endgroup$
    – Sam Nead
    Oct 31, 2022 at 21:14
  • 2
    $\begingroup$ When I say "Waldhausen proves" I mean "I only vaguely remember this". Looking at the reference you suggest, in fact Corollary 6.5 is exactly what you want. Yes, boundary irreducible allows empty boundary. $\endgroup$
    – Sam Nead
    Oct 31, 2022 at 21:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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