MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $\Gamma\subset {SL}_2(\mathbb Z)$ be a congruence subgroup, and let $X(\Gamma)$ be the associated compact modular curve. The inclusion $\Gamma \subset SL_2(\mathbb Z)$ induces a canonical projection $p:X(\Gamma)\to X(1)$ which is a non-constant holomorphic map of compact Riemann surfaces, where $X(1)=X(SL_2(\mathbb Z))$. The map $p$ ramifies at most over the elliptic points $i=\sqrt{-1}$ and $\mu_3=e^{2\pi i/3}$ of $X(1)$ and over the cusp $\infty$ of $X(1)$. Let $d$ be the degree of $p$ and let $\overline{\mathbb Q}$ be an algebraic closure of $\mathbb Q$.

Assumption $(*)$: There exists a smooth, projective and connected curve $X(\Gamma)_{\overline{\mathbb Q}}$ over $\overline{\mathbb Q}$ such that its base change to $\mathbb C$ corresponds to $X(\Gamma)$.

  1. If assumption $(*)$ is satisfied, does there exists a non-constant morphism $X(\Gamma)_{\overline{\mathbb Q}}\to \mathbb P^1$ of algebraic curves over $\bar{\mathbb Q}$ of degree $d$, which ramifies at most over $i$, $\mu_3$ and $\infty$ (viewed as points of $\mathbb P^1=X(1)_{\overline{\mathbb Q}}$)?

  2. Is assumption $(*)$ always satisfied?

share|cite|improve this question
Yes and yes. You don't deduce 2 from 1, you prove them both at the same time by proving that if $\Gamma=\Gamma(N)$ is the principal congruence subgroup, then there's a non-compactified moduli space representing elliptic curves over $\overline{\mathbf{Q}}$ with two points of order $N$ generating the $N$-torsion; then you let $Y(\Gamma)$ be one of the components of this moduli space and let $X(\Gamma)$ be the compactification. Both (1) and (2) now follow; to deal with general $\Gamma$ you just quotient out $X(\Gamma(N))$ by the finite group $\Gamma/\Gamma(N)$. Done! – user30035 Mar 14 '13 at 22:45
meh should say "don't deduce 1 from 2". Roll on the days when I can edit comments. – user30035 Mar 14 '13 at 22:46
Dear wccanard, thanks for the answer. – Guillaume Pastorini Mar 15 '13 at 8:50
up vote 1 down vote accepted

Let $X(\Gamma)\to \mathbf{P}^1_{\mathbf C}$ be the composition of the natural map $X(\Gamma)\to X(1)$ associated to the inclusion $\Gamma\subset$ SL$_2(\mathbf Z)$, and the $j$-invariant $j:X(1)\to \mathbf{P}^1_{\mathbf C}$. This map is, as you said, ramified over at most three points. Since these three points are algebraic numbers, by a theorem of Grothendieck and Weil, there exists a unique pair $(Y,Y\to \mathbf{P}^1_{\overline{\mathbf Q}})$ with $Y$ a smooth projective connected curve over $\overline{\mathbf Q}$ and $Y\to \mathbf{P}^1_{\overline{\mathbf Q}}$ a finite morphism which gives the natural map $X(\Gamma)\to \mathbf{P}^1_{\mathbf C}$ after base-change. If you just want a model for your $X(\Gamma)$ over $\overline {\mathbf Q}$ as a curve (and not as a cover) you have more choice.

The theorem of Grothendieck and Weil is more general. Let $k\subset K$ be an extension of algebraically closed fields of characteristic zero. Then, for any smooth quasi-projective connected variety $U$ over $k$, the base-change functor is an equivalence of categories from "finite etale covers of $U$" to "finite etale covers of $U_K$". We are applying this theorem to $\mathbf P^1_{k} -\{0,1,\infty\}$ with $k=\overline {\mathbf Q}$ and $K=\mathbf C$.

share|cite|improve this answer
Dear Ariyan Javanpeykar. Thank you very much for your excellent answer. Just to make sure that I understand your arguments completely: 1. The non-constant holomorphic map $X(\Gamma)\to X(1)$ of compact Riemann surfaces induces a non-constant morphism $X(\Gamma)\to X(1)$ of algebraic curves over $\mathbb C$, with the same degree and the same ramification points? 2. What is the etale cover $V\to U=\mathbb P^1-\{0,1,\infty\}$ to which you are applaying G-W (is $V=(X(\Gamma)-\{points \ above \ 0,1,\infty\})$ and then you restrict the natural map)? – Guillaume Pastorini Mar 15 '13 at 9:28
1. Yes. 2. V is the inverse image of U as you already suspected. Let me know if anything else seems obscure. – Ariyan Javanpeykar Mar 16 '13 at 20:40
This answer is certainly correct, but wccanard makes an important point: the viewpoint of the original question may be "mistaken" (depending on the motivation). The relationship between certain maps among algebro-geometric moduli spaces and maps between their "uniformized analytifications" rests on arguments that have a lot to do with GAGA and moduli functors (and nothing to do with Grothendieck-Weil). For many applications in number theory (work of Mazur, Gross-Zagier, etc.), one needs this alternative viewpoint in order to work with points over specific number fields, etc. – user30379 Mar 17 '13 at 5:48
@pranavk You are absolutely right. The point wccanard makes is certainly important. The above point of view doesn't take any moduli interpretation into account which Guillaume probably wants. I recommend Guillaume to keep my answer in mind (also when dealing with certain higher-dimensional Shimura varieties), but to always try and use the moduli-interpretation as much as possible as wccanard does. It's more natural when working with modular curves. – Ariyan Javanpeykar Mar 17 '13 at 15:03

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.