Let $Y$ be a hyperbolic manifold that fibers over $S^1$, with fibration $\pi:Y \to S^1$ with fiber $\Sigma$. Thurston states that the monodromy $\phi:\Sigma \to \Sigma$ of this projection is then pseudo-Anosov. Moreover, we can write $Y = N/\mathbb{Z}$ where $N \to \mathbb{R}$ is a hyperbolic manifold fivering over $\mathbb{R}$ with fiber $\Sigma$ and $\mathbb{Z}$ acts on $N$ by isometries. Let $\Phi$ be the generator of the $\mathbb{Z}$-action.

Question. Is there a way of constructing a canonical smooth respresentative of the monodromy $\phi$ of the fiber bundle using the deck transformation $\Phi$?

For instance, you could imagine that there is a projection $N \to \Sigma$ (perhaps with geodesic fibers) where $\Phi$ permutes the fibers of the projection, and where $\Phi$ descends to a diffeomorphism of $\Sigma$ isotopic to $\phi$.

I'm particularly interested in the case of a fibered hyperbolic knot complement.

  • 1
    $\begingroup$ Do you know what "pseudo-Anosov" means? $\endgroup$ Mar 23, 2023 at 20:38
  • 4
    $\begingroup$ @MoisheKohan Yes. If the answer is obvious to you, feel free to share. $\endgroup$ Mar 23, 2023 at 21:34
  • $\begingroup$ Oh, I missed that you require smoothness. Then you can use a harmonic map or use conformal barycentric extension. $\endgroup$ Mar 24, 2023 at 1:05
  • 2
    $\begingroup$ What would "canonical" mean to you? It's a difficult word in this particular context: you may know that the pseudo-Anosov homeomorphism $\phi$ itself is canonical in a topological sense (it is unique up to conjugacy by a homeomorphism isotopic to the identity), but it generally fails to be smooth. $\endgroup$
    – Lee Mosher
    Mar 24, 2023 at 15:48

2 Answers 2


Here is an answer of sorts; it is not completely canonical though. First of all, you have to pick a conformal or hyperbolic structure on the fiber $\Sigma$. This can be made almost canonical, since there is a unique $\phi$-invariant (real) Teichmuller geodesic. In any case, you get $\Sigma={\mathbb H}^2/\Gamma$, where $\Gamma$ is $\pi_1(\Sigma)$ acting on the hyperbolic plane (isometrically, freely, properly). The map $\phi$ induces a canonical $\Gamma$-equivariant homeomorphism $f: S^1\to S^1$ (of the boundary circle of the hyperbolic plane). Now, use the Douady-Earle extension to extend $f$ to a $\Gamma$-equivariant diffeomorphism $F$ of the hyperbolic plane. Lastly, project $F$ to a diffeomorphism $\phi$ of the surface $\Sigma$.

  • $\begingroup$ This is an excellent answer. A few follow up questions (1) is the homeomorphism of the ideal boundary of $\mathbb{H}$ determined canonically by the mapping class? Or are you suggesting to first take the usual non-smooth representative of $\phi$ and then to extend its action on the boundary? $\endgroup$ Mar 27, 2023 at 17:50
  • $\begingroup$ (2) Is it possible to make the map area preserving? I'm not sure if this is a property of the Douady-Earle extension. I noticed this paper: arxiv.org/abs/0911.4124 which gives a different extension with this property. $\endgroup$ Mar 27, 2023 at 17:59
  • $\begingroup$ (3) Also any basic references on this material would be amazing :) I'm very much a novice in these things :) Thank you! $\endgroup$ Mar 27, 2023 at 18:00
  • $\begingroup$ Check "Primer on mapping class group" by Farb and Margalit. $\endgroup$ Mar 27, 2023 at 19:29
  • $\begingroup$ I think DE extension is area preserving. $\endgroup$ Mar 27, 2023 at 19:31

The pseudo-Anosov homeomorphism is a diffeomorphism away from finitely many singular points, so in general will not be a smooth diffeomorphism. However, for certain fibered knots there are not singularities away from the punctures, and hence the map is a smooth diffeomorphism and unique up to smooth conjugacy. This works for the figure 8 knot and more generally certain 2-bridge knots (considered in this paper by Sakata ).

In general though there will be interior singularities of the pseudo-Anosov map of the fiber. Gerber and Katok proved that a pseudo-Anosov map is topologically conjugate to a smooth diffeomorphism. So this gives a kind of canonical smooth diffeomorphism realization, but only up to topological conjugacy (I do not know if diffeomorphism realizations are smoothly conjugate; there are also analytic realizations).


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.