Loading [MathJax]/extensions/TeX/boldsymbol.js
Skip to main content

Section 2.4 Riemann Surfaces and Discrete Groups (Rod)

Welcome to BUGLES (Boston university geometry learning expository seminar), the reason it is called bugles is because bugles are hyperbolic, and today we will see a lot of hyperbolic objects.

Plan

  1. Uniformization
  2. Fuchsian groups
  3. Automorphisms of Riemann surfaces

Subsection 2.4.1 Uniformization

\(g = 0\) Uniformization.

\(g \ge 1\) \(\hat \CC\) can't be a cover by Riemann-Hurwitz. \(g = 1\) \(\pi_1 (\Sigma) = \ZZ \oplus \ZZ\) abelian.

Claim: no subgroup of \(\Aut (\HH) \) is isomorphic to \(\ZZ \oplus \ZZ\) acting freely and properly discontinuously. So \(\widetilde\Sigma = \hat \CC\) \(z\mapsto az+b\) free id \(a=1\) so \(z\mapsto z+\lambda_1\) \(z\mapsto z+\lambda_2\text{.}\)

\(g= 2\) \(\pi_1(\Sigma)\) is not abelian but \(z\mapsto z + \lambda_1\) is abelian!

\begin{equation*} \Sigma = \HH/K,\, K \subseteq \PSL_2(\RR)\text{.} \end{equation*}
Goal.

Understand \Sigma through \widetilde \Sigma and G\text{.}

Fuchsian groups.

g \ge 2\text{.}

\begin{equation*} \Aut (\HH) = \PSL_2(\RR) = \operatorname{Isom}^+ ( \HH, \frac{|\diff z|^2}{\Im Z}) \end{equation*}

hyperbolic \HH, \mathbf D and \PSL_2(\RR) acts transitively on geodesics.

Definition 2.4.5. Fuchsian groups.

A Fuchsian group is a discrete subgroup of \PSL_2(\RR)\text{.}

Remark 2.4.6.

(proof in book) Even if \Gamma doesn't act freely the quotient

\begin{equation*} \HH \to \HH/\Gamma \end{equation*}

is still a covering map and \HH/\Gamma is a Riemann surface.

Reflections on \HH.

Say \mu is a geodesic in \HH\text{,} i.e. a horocycle. There is M \in \PSL_2(\RR) with M\mu the imaginary axis. Then R = -\bar z is the reflection over the imaginary axis. Now R_\mu = M\inv\circ R \circ M is a reflection over \mu\text{.}

\begin{equation*} R_\mu = \frac{a\bar z + b}{ c\bar z + d}\not\in \PSL_2(\RR) \end{equation*}

this is a a problem for us.

Triangle groups.

Given n, m, l\in \ZZ \cup \{\infty\} then there is a hyperbolic triangle with angles \pi/n,\pi/m, \pi/l if

\begin{equation*} \frac 1n + \frac 1m + \frac 1l \lt 1\text{.} \end{equation*}

With area \pi(1 - \frac 1n - \frac 1m - \frac 1l)\text{.}

In the disk model we can start with a wedge of the disk and by adding a choice third geodesic with endpoints on the edge we can adjust the other angles to be what we like.So we can construct hyperbolic triangles with whatever angles we like. Then let R_1 be the reflection over 1 edge, R_2 , R_3 similarly. By reflecting our original triangle T with these reflections we can tessellate the disk, colouring alternately the triangles obtained using an odd or even number of reflections.

The only remaining problem is that R_i's are not in \PSL_2(\RR)\text{.} The solution is to define x_1 = R_3 \circ R_1\text{,} x_2 = R_1 \circ R_2\text{,} x_3 = R_2 \circ R_3 which are all in \PSL_2(\RR) now. Now we need to take the union of two adjacent triangles before as a fundamental domain, some quadrilateral that still tessellates. So we have formed a Fuchsian group from our triangles.

A presentation for this group is

\begin{equation*} \langle x_1, x_2, x_3 | x_1 ^n = x_2 ^ m = x_3 ^l = x_1x_2x_3 = 1\rangle \end{equation*}

note n,m,l can still be \infty\text{.}

Definition 2.4.7. Triangle groups.

Let \Gamma_{n,m,l} be the triangle group with signature (1/n, 1/m, 1/l)\text{.}

Remark 2.4.8.
\begin{equation*} \frac 1n + \frac 1m + \frac 1l = 1 \end{equation*}
\begin{equation*} \frac 1n + \frac 1m + \frac 1l \gt 1 \end{equation*}

still work on \CC and \hat \CC respectively.

Example 2.4.9. \PSL_2(\ZZ).

Consider \Gamma_{2,3,\infty} angles \pi/2, \pi/3, 0\text{.} We can draw such a triangle in the upper half plane with vertices i, e^{\pi i/3}, \infty\text{.} So a fundamental domain will be the region obtained by reflecting through the imaginary axis, given by -\frac 12 \le \Re z \le \frac 12\text{,} |z| \ge 1\text{.} We have R_1 = \frac{1}{\bar z}, R_2 = -\bar z +1,R_3 = -\bar z so x_1 = \frac{-1}{z}, x_2 = \frac{1}{-z+1}, x_3 = z+1\text{.} Then \Gamma_{2,3,\infty} \cong \PSL_2(\ZZ)\text{.} Sometimes denoted \Gamma(1)\text{.}

Observation 2.4.10.

If \Gamma_1 \lt \Gamma_2 and T is a fundamental domain of \Gamma_2 then if \gamma_1, \gamma_2, \ldots, \gamma_n \in \Gamma_2 are representatives of \Gamma_1\backslash \Gamma_2 then

\begin{equation*} \bigcup \gamma_i (T) \end{equation*}

is a fundamental domain for \Gamma_1\text{.}

Example 2.4.11. \Gamma(1).
\begin{equation*} \Gamma(2) = \{\begin{pmatrix} a\amp b \\ c\amp d\end{pmatrix} = \id \pmod 2 \} \end{equation*}

then

\begin{equation*} [\Gamma(1) : \Gamma(2)] = 6 \end{equation*}

representatives of \Gamma(2) \backslash \Gamma(1) are

\begin{equation*} x_1 = \id,\,x_2 = \frac{-1}{z-1}, x_3 = \frac{z-1}{z},\, x_4 = \frac{z-1}{z},\,x_5 = \frac{-z}{x-1},\,x_6 = \frac{-1}{z}\text{.} \end{equation*}

Lets see what these do, for example if z= e^{i\theta}

\begin{equation*} \Re(x_2(z) = \frac{-1}{e^{i\theta} - 1} = \frac{-e^{i\theta} + 1}{2- 2\cos \theta}) = \frac{1-\cos \theta}{ 2- 2\cos \theta} \frac 12 \end{equation*}

if we plot this we see we get two copies of a 0,0,0 triangle so this corresponds to \Gamma_{\infty,\infty,\infty}\text{.}

\begin{equation*} \langle x_1, x_2, x_3 | x_1x_2x_3 = 1\rangle = \langle x_1,x_2\rangle = \pi_1(\PP^1\smallsetminus\{0,1,\infty\})\text{.} \end{equation*}

\(\Leftarrow\) Define an \(\phi\colon S_1 \to S_2\) via \(\phi(\lb z\rb_1) = \lb T(z)\rb_2\text{.}\)

\(\Rightarrow\) Take a lift

\begin{equation*} \xymatrix{ \HH\ar[r]^{\tilde\phi}\ar[d] & \HH\ar[d] \\ \HH/\Gamma_1\ar[r]_\phi & \HH/\Gamma_2 } \end{equation*}

then \(T= \tilde \phi\text{.}\)

Previous proposition, set \(\Gamma_1 = \Gamma_2\)

\begin{equation*} N(\Gamma) \to \Aut(\HH/\Gamma) \end{equation*}

kernel is \(\Gamma\text{.}\)

\begin{equation*} \xymatrix{ \HH \ar[dr]^{\phi_1} \ar[d]_{\phi_2}\\ S =\HH/\Gamma \ar[r]_f & \HH/N(\Gamma) = S/\Aut(S) } \end{equation*}

since \(\phi_1,\phi_2\) are holomorphic then so is \(f\text{.}\) So \(\deg f= | N(\Gamma) /\Gamma|\) and \(\deg f \lt \infty\text{.}\)

Say \Sigma\text{,} g\ge 2\text{,} G\subseteq \Aut(\Sigma)\text{.} Let \bar g be the genus of \Sigma /G

\begin{equation*} 2g - 2 = |G| (2\bar g - 2) + \sum_p (I(p) - 1) = |G|(2\bar g - 2 + \sum_{i=1}^n (1 - \frac{1}{|I(p_i)|})) \end{equation*}

where I(p) is the stabiliser of p in G and \{p_i\} area maximal set of fixed points of G inequivalent under the action of G\text{.}

\(\Sigma,\,g\ge 2\) then \(|\Aut(\Sigma)| \le 84(g-1)\text{.}\) Hint: cases.

Consider

\begin{equation*} 1 \to \Gamma(n) \to \Gamma(1) \to \PSL_2(\ZZ/n\ZZ) \to 1 \end{equation*}

compute genus of \(\HH/\Gamma(n)\text{.}\)