If $f$ is an orientation preserving homeomorphism from $S^1$ to $S^1$ and $\tau$ is a tessellation of $\mathbb{D}$ (that is a collection of locally finite geodesics decomposing $\mathbb{D}$ into ideal triangles) then $f(\tau)$ is also a triangulation.

(Locally finite: Any point of $\mathbb{D}$ admits a neighborhood meeting only finitely many geodesics in $\tau$. )

I have read this statement many times but haven't seen any proof of it. Can anyone please give a complete proof of this. In particular I want to know why the image will be locally finite.


1 Answer 1


Given a geodesic $\gamma$, let $r(\gamma)$ be the Euclidean distance from the center of the disk to $\gamma$.

Claim 1. A collection $\Gamma$ of geodesics is locally finite if and only if for every $\epsilon>0$ the set $\{\gamma\in\Gamma: r(\gamma)\le 1-\epsilon\} $ is finite.

Proof: If $\Gamma$ is locally finite, then only finitely many geodesics meet the closed Euclidean disk $\{z:|z|\le 1-\epsilon\}$, by compactness of the latter. In the converse direction, every point of $ \mathbb D$ has a neighborhood contained in some set of the form $\{z:|z|\le 1-\epsilon\}$. $\Box$

Let $d (\gamma)$ be the Euclidean distance between the endpoints of $\gamma$.

Claim 2. There is a continuous strictly decreasing function $\phi:[0,1)\to (0,2]$ such that $d(\gamma)=\phi(r(\gamma))$ for every geodesic. Also, $\phi(1-)=0$.

Proof: either of two quantities $d$ and $r$ determines the geodesic up to Euclidean rotation, and therefore determines the other quantity. A rough upper bound for $\phi$ is $\phi(r)\le 2\sqrt{1-r^2} $, which suffices for $\phi(1-)=0$. This bound follows from the fact that a Euclidean line passing at distance $r$ from the origin cuts off a part of the disk with Euclidean diameter $2\sqrt{1-r^2}$. $\Box$

The function $\phi$ could be written down explicitly, but this is not necessary.

The image of a geodesic $\gamma$ under a map $f:S^1\to S^1$ is understood as the geodesic between the images of the endpoints of $\gamma$.

Claim 3. if $\tau$ is a locally finite collection of geodesics, and $f:S^1\to S^1$ is continuous, then $f(\tau)$ is locally finite.

Proof: Given $\epsilon>0$, we must show that $\{ \gamma \in\tau : r(f(\gamma))\le 1-\epsilon\}$ is finite. By the uniform continuity of $f$, there is $\delta>0$ such that $d(f(\gamma))<\phi(1-\epsilon)$ whenever $d(\gamma)<\delta$. Since $\phi(1-)=0$, there is $\tau>0$ such that $d(\gamma)<\delta$ whenever $r(\gamma)>1-\tau$.

The chain of containments $$\begin{split}\{ \gamma \in\tau : r(f(\gamma))\le 1-\epsilon\} &= \{ \gamma \in\tau : d(f(\gamma)) \ge \phi(1-\epsilon)\} \\ &\subseteq \{ \gamma \in\tau : d( \gamma ) \ge \delta\} \\ &\subseteq \{ \gamma \in\tau : r( \gamma ) \le 1-\tau\} \end{split}$$ proves the claim. $\Box$

This takes care of local finiteness. Another thing to check is that $f(\tau)$ is a tesselation when $\tau$ is a tesselation and $f$ is a homeomorphism. First, the endpoints of geodesics in $f(\tau)$ are dense in $S^1$, because the image of a dense set is dense. Second, the geodesics in $f(\tau)$ do not intersect. Indeed, two geodesics $\gamma_1,\gamma_2$ intersect if and only if the endpoints of $\gamma_1$ separate the endpoints of $\gamma_2$ in $S^1$ (in other words, removing the endpoints of one geodesic leaves the endpoints of the other in different connected components). This property is also invariant under homeomorphisms.


You must log in to answer this question.

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