3
$\begingroup$

Let $\tau$ be a CM point and let $\mathcal O$ be the quadratic order corresponding to the lattice $[\tau,1]$, that is $$\mathcal O =\lbrace \lambda \in \mathbb C: \lambda[\tau,1]\subset[\tau,1]\rbrace.$$

Embed $\mathcal O$ in $2$-by-$2$ matrices via its action on $[\tau,1]$: $$f\colon \mathcal O\longrightarrow \operatorname{M}_2(\mathbb Z).$$

Let $N$ be a positive integer. The embedding $f$ induces a homomorphism $$\mathcal O/N\mathcal O \longrightarrow \operatorname{M}_2(\mathbb Z/N\mathbb Z),$$ $$a+N\mathcal O\longmapsto f(a)+N\cdot\operatorname{M}_2(\mathbb Z).$$

Is this homomorphism injective?

$\endgroup$

1 Answer 1

3
$\begingroup$

Yes, it is injective. Write $\Lambda$ in place of $[\tau,1]$. Let $a$ be in the kernel. This means that the matrix of $a$ is $0$ modulo $N$ or, in other words, $a \Lambda \subseteq N \Lambda$. But then $(a/N) \Lambda \subseteq \Lambda$ so, by definition of $\mathcal{O}$, we have $a/N \in \mathcal{O}$. This shows $a \in N \mathcal{O}$ and $a$ is $0$ in $\mathcal{O}/N \mathcal{O}$, as desired.

$\endgroup$

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.