For a positive integer $N$, we define $$\Gamma(N)=\big \{ \begin{bmatrix} a & b \newline c & d\end{bmatrix}\in \operatorname{SL}_2(\mathbb{Z}): \begin{bmatrix} a & b \newline c & d\end{bmatrix}\equiv \begin{bmatrix} 1 & 0 \newline 0 & 1\end{bmatrix}(\operatorname{mod}N)\big\}$$

Then, [Diamond, Shurman] defines the set of equivalence classes $S(N)$ of enhanced elliptic curves associated to $\Gamma(N)$ to be the set of equivalence classes of all triples $(E,P,Q)$, where $E/\mathbb{C}$ is an elliptic curve, and $P,Q\in E(\mathbb{C})[N]$ such that the Weil pairing $e_N(P,Q)=e^{2\pi i/N}$ .

Defining a Weil pairing:

Let $P_1,P_2\in E(\mathbb{C})[N]$ (possibly equal). Firstly, fix $E(\mathbb{C})=\mathbb{C}/\Lambda$ for some lattice $\Lambda=\mathbb{Z}\omega_1+\mathbb{Z}\omega_2\subset \mathbb{C}$. Now, there is a matrix $\gamma\in M_{2\times 2}(\mathbb{Z}/N\mathbb{Z})$ such that $\begin{bmatrix} P_1 \\ P_2\end{bmatrix}=\gamma \begin{bmatrix} \omega_1/N+\Lambda \\ \omega_2/N+\Lambda\end{bmatrix}$. We define $e_N(P_1,P_2)=e^{2\pi i\operatorname{det}(\gamma)/N}$.

Now, I know that there exists a smooth affine curve $X_{\widetilde{\Gamma}(N)}/\operatorname{Spec}\mathbb{Z}[1/N]$ which represents the moduli problem (assume $N$ is large enough): $$X_{\widetilde{\Gamma}(N)}(S)=\{(E,\alpha):E/S \operatorname{\ is\ an\ elliptic\ curve\ }, \alpha:\underline{(\mathbb{Z}/N\mathbb{Z})^{\oplus 2}}_S\xrightarrow{\cong}E[N]\}.$$

At the face of it, it seems like $X_{\widetilde{\Gamma}(N)}(\mathbb{C})=S(N)$. But, the problem I am having is that an isomorphism $\alpha:(\mathbb{Z}/N\mathbb{Z})^{\oplus 2}\xrightarrow{\cong}E(\mathbb{C})[N]$ gives two points $P,Q\in E(\mathbb{C})[N]$ such that $e_N(P,Q)=e^{2\pi i x/N}$ for some $x\in (\mathbb{Z}/N\mathbb{Z})^\times$, which is not the same as a point in $S(N)$

Am I missing something? Are these two the same?


Diamond, Fred; Shurman, Jerry, A first course in modular forms, Graduate Texts in Mathematics 228. Berlin: Springer (ISBN 0-387-23229-X/hbk). xv, 436 p. (2005). ZBL1062.11022.


1 Answer 1


This is a subtle issue (which has come up before on this site several times, see e.g. is the modular curve X(N) defined over Q? for a related question).

Your $S(N)$ is naturally a scheme over $\mathbb{Z}[1/N, \zeta_N]$. Your $X_{\widetilde{\Gamma}(N)}$ is, as you have defined it, a scheme over $\mathbb{Z}[1/N]$. But there is actually a natural map $\mathbb{Z}[1/N, \zeta_N] \hookrightarrow \mathcal{O}(X_{\widetilde{\Gamma}(N)})$ which sends $\zeta_N$ to $e_N(P, Q)$; and if you regard $X_{\widetilde{\Gamma}(N)}$ as a $\mathbb{Z}[1/N, \zeta_N]$-scheme via this morphism, it coincides with $S(N)$.

  • $\begingroup$ When you think of $e_N(P,Q)\in \mathcal{O}(X_{\widetilde{\Gamma}(N)})$, what is $P$ and $Q$? Are these some fixed generators of the $N$-torsion points of the universal elliptic curve? $\endgroup$ Commented May 9, 2023 at 21:26
  • $\begingroup$ They are the canonical $N$-torsion points of the universal elliptic curve (which are part of what it means for the space to represent a functor). $\endgroup$ Commented May 10, 2023 at 0:05
  • $\begingroup$ I had two follow-up question. Firstly, to summarise what you have said, would it be correct to say that over $\mathbb{Z}[1/N,\zeta_N]$, $X_{\widetilde{\Gamma}(N)}(\mathbb{C})=\Gamma(N)\backslash \mathcal{H}$, but over $\mathbb{Z}[1/N]$ this is not true anymore? $\endgroup$ Commented May 10, 2023 at 7:18
  • $\begingroup$ Secondly, over $\mathbb{Z}[1/N$], will we get $X_{\widetilde{\Gamma}(N)}(\mathbb{C})=\sqcup_{(\mathbb{Z}/N\mathbb{Z})^{\times}}\Gamma(N)\backslash \mathcal{H}$ $\endgroup$ Commented May 10, 2023 at 7:21
  • 1
    $\begingroup$ Perhaps the same thing in better language: this is about whether we study maximal ideals of $O(X) \otimes_{\mathbb{Z}[1/N, \zeta_N]}\mathbb{C}$ or of $O(X) \otimes_{\mathbb{Z}[1/N]} \mathbb{C}$. The $\mathbb{C}$-points of the first are $\Gamma(N) \backslash \mathcal{H}$. The $\mathbb{C}$-points of the second are $\varphi(N)$ copies of that. ($X = X_{\tilde{\Gamma}(N)}$) $\endgroup$ Commented May 10, 2023 at 14:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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