arXiv2019
This work is a contribution to the classification of Teichmüller curves in the moduli space $\M_2$ of Riemann surfaces of genus 2. While the classification of primitive Teichmüller curves in $\M_2$ is complete, the classification of the imprimitive curves, which is related to branched torus covers and square-tiled surfaces, remains open. Conjecturally, the classification is completed as follows. Let $W_{d^2}[n] \subset \M_2$ be the 1-dimensional subvariety consisting of those $X \in \M_2$ that admit a primitive degree $d$ holomorphic map $π: X \to E$ to an elliptic curve $E$, branched over torsion points of order $n$. It is known that every imprimitive Teichmüller curve in $\M_2$ is a component of some $W_{d^2}[n]$. The {\em parity conjecture} states that (with minor exceptions) $W_{d^2}[n]$ has two components when $n$ is odd, and one when $n$ is even. In particular, the number of components of $W_{d^2}[n]$ does not depend on $d$. In this work we establish the parity conjecture in the following three cases: (1) for all $n$ when $d=2,3,4,5$; (2) when $d$ and $n$ are prime and $n > (d^3-d)/4$; and (3) when $d$ is prime and $n > C_d$, where $C_d$ is an implicit constant that depends on $d$. In the course of the proof we will see that the modular curve $X(d) = \overline{\Hyp \big/ Γ(d)}$ is itself a square-tiled surface equipped with a natural action of $\SLZ$. The parity conjecture is equivalent to the classification of the finite orbits of this action. It is also closely related to the following {\em illumination conjecture}: light sources at the cusps of the modular curve illuminate all of $X(d)$, except possibly some vertices of the square-tiling. Our results show that the illumination conjecture is true for $d \le 5$.