Let $S_{g,1}$ be the surface of genus $g \geq 1$ and $1$ boundary component. Let $Mod(S_{g,1})$ be the mapping class group in which we allow isotopies to rotate the action on the boundary (equivalently think of it as the mapping class group of the once-punctured surface of genus $g$).

Is every element of $Mod(S_{g,1})$ a composition of right-handed Dehn twists?

Note that this is true for $S_{g,0}$ as stated in page 124 of A primer on Mapping Class Groups by Farb and Margalit under the name of "a strange fact".

Edit: I am going to comment ThiKu's answer to avoid further confusion.

In A. Wand: Factorisation of Surface Diffeomorphisms and in Baker, Etnyre and Van Horn-Morris: Cabling, Contact Structures and and Mapping Class Monoids the authors, independently, provide with examples of diffeomorphisms in $Veer(\Sigma_{2,1},\partial \Sigma_{2,1})$ which are not in $Dehn^+(\Sigma_{2,1}, \partial \Sigma_{2,1})$. That is, right-veering diffeomorphisms which are not a product of right-handed Dehn twists. However, the mapping class group in which these results hold is $Mod( \Sigma_{2,1}, \partial \Sigma_{2,1})$, that is, the mapping class group of automorphisms fixing the boundary and isotopies fixing the boundary as well. This, a priori, does not yield counter-examples to my question (unless it does together with some other result that I do not know).

Observe that for all $g \geq 1$ there is a central extension

$$1 \to \mathbb{Z} \to Mod(\Sigma_{g,1}, \partial \Sigma_{g,1}) \to Mod( \Sigma_{g,1}) \to 1 $$

which is not split in general.


No. A. Wand: Factorisation of Surface Diffeomorphisms discusses conditions when a mapping class is a product of right-handed Dehn twists. Theorem 5.2 of that paper gives an explicit counterexample to your question.

  • $\begingroup$ I deleted my previous comment because now I think these are not counter examples. Note that I asked for the mapping class group free on the boundary. It could happen that the image of Wands examples in $Mod(S_{g,1})$ are a composition of right handed Dehn twists. And that, by taking this composition in $Mod(S_{g,1}, \partial S_{g,1})$ you get the original automorphism plus some right handed Dehn twists around the boundary parallel curve. $\endgroup$
    – Paul
    Oct 29 '18 at 14:39
  • $\begingroup$ Actually the identity is a (non-empty) composition of right Dehn twists in the mapping class group free on the boundary but not in the mapping class group relative on the boundary $\endgroup$
    – Paul
    Oct 29 '18 at 14:45
  • 3
    $\begingroup$ @Paul: I think you've answered you're own question. If the identity is a product of right Dehn twists, then the whole mapping class group is. Take the product of right twists, set it $=1$, and move one twist to the right hand side of the equation. This expresses the inverse of right-hand twist (a left-hand twist) as a product of right-hand twists. Since the set of all Dehn twists generate, you can generate the mapping class group with right-hand twists. This is stated in the "strange fact" on p. 124 of Farb-Margalit. $\endgroup$
    – Ian Agol
    Oct 29 '18 at 19:33
  • $\begingroup$ I had in mind a particular example for g=1 that's why I did not answer... but now I think that, since all genus and boundary components combinations appear as milnor fibers of brieskorn-pham singularities on $C^2$, the result is true in greater generality. I am going to think a little bit about it and if it is not true, at least I will say that it is true for some genus. $\endgroup$
    – Paul
    Oct 29 '18 at 19:51

Yes. As pointed out by Ian Agol, I actually answered my question.

I observed that the identity is a non-empty composition of right handed Dehn twists in $Mod(S_{g,1})$. A priori this is not trivial. I was thinking about monodromies on Brieskorn-Pham singularities $(x^p+y^q)$ which are freely periodic and a composition of right-handed Dehn twists (by morsifying the singularity). One can easily see that this solves the problem in the cases that I was asking originally since all surfaces $S_{g,1}$ appear as Milnor fibers of such singularities for all $g$.

EDIT: I removed the last part of the answer since it was not true in that generality and this answers completely what I wanted.


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.