3
$\begingroup$

Let $\Gamma \subset \mathrm{SL}(2,\mathbb{R})$ be a lattice. If $N_1, N_2$ are a pair of independent parabolic subgroups contained in $\Gamma$, why must $\Gamma$ contain a hyperbolic element? By parabolic subgroup, I mean "subgroup containing only parabolic elements, other than the identity." This is used in the proof of Theorem 10.1, here:

https://people.math.harvard.edu/~ctm/papers/home/text/papers/abel/abel.pdf

If it helps, $\Gamma$ here is the stabilizer of a 1-form defining a Teichmuller curve.

$\endgroup$
1
  • 2
    $\begingroup$ The point is that, and this is true in any Gromov-hyperbolic space, the product of two (high powers of) independent parabolic isometries is loxodromic. $\endgroup$
    – AGenevois
    Commented Aug 9, 2022 at 5:32

1 Answer 1

4
$\begingroup$

This very special case follows from "general principles" (namely a version of the ping-pong lemma) but it is also possible to give a direct proof, as follows.


Suppose that $a$ and $b$ are the distinct points at infinity fixed by the two parabolic subgroups $A$ and $B$ (note change of notation). Since $\mathrm{SL}(2, \mathbb{R})$ is three-transitive, we can conjugate $\Gamma$ and so assume that $a = \infty$ and $b = 0$. We deduce that elements of $A$ now have the form $$ \begin{pmatrix} 1 & r\\ 0 & 1 \end{pmatrix} $$ while elements of $B$ have the form $$ \begin{pmatrix} 1 & 0\\ s & 1 \end{pmatrix} $$ Taking an inverse if needed, we obtain such elements of $A$ and $B$ where $r$ and $s$ are non-zero and have the same sign. We now multiply to find $$ \begin{pmatrix} 1 & r\\ 0 & 1 \end{pmatrix} \begin{pmatrix} 1 & 0\\ s & 1 \end{pmatrix} = \begin{pmatrix} 1 + rs & r\\ s & 1 \end{pmatrix} $$ This has trace $2 + rs > 2$ so is hyperbolic, as desired.


Hmm. Now that I write this, I realise that there is a third proof using the classification of fixed points of isometries, and the intermediate value theorem. It helps to draw a picture and think dynamically.

$\endgroup$
6
  • $\begingroup$ Great, I was wanting something like this, as pseudo-Anosov diffeomorphisms have derivatives with trace larger than 2, and you can produce any one of these (Thurston) from a product of parabolic elements $\endgroup$ Commented Aug 9, 2022 at 13:23
  • $\begingroup$ It is not true that all pA maps arise from the Thurston-Veech construction. See arxiv.org/abs/1410.6974 $\endgroup$
    – Sam Nead
    Commented Aug 9, 2022 at 18:34
  • $\begingroup$ I see. But you can produce pA diffeomorphisms from a product of two transverse parabolic elements, using the Thurston-Veech construction, correct ? $\endgroup$ Commented Aug 9, 2022 at 19:59
  • $\begingroup$ Yes, some pA maps arise in this way. In particular, if you have a flat surface that decomposes nicely (in two ways) as a union of annuli with rationally related moduli, then the right twist (aka shear) on one, followed by the left twist on the other, will be pA. The Thurston--Veech construction is usually phrased as a special case of this, where the flat surface is tiled by squares. $\endgroup$
    – Sam Nead
    Commented Aug 9, 2022 at 20:10
  • $\begingroup$ Is it possible to give a similarly elementary argument to prove the same statement in dim>2? $\endgroup$
    – Dinisaur
    Commented Feb 24 at 18:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .