2
$\begingroup$

Let $E/\mathbb{Q}$ be an elliptic curve and let $p$ be a prime. Then there is an action of the absolute Galois group of $\mathbb{Q}$ on $E[p]$ that factors through a finite quotient.

Does any finite quotient of the absolute Galois group of $\mathbb{Q}$ arise this way for some $E$ and $p$? What if we consider abelian varieties instead of elliptic curves?

$\endgroup$
3
  • 3
    $\begingroup$ Every such action is odd, meaning the complex conjugation is mapped to multiplication by $-1$ (this is certainly true for elliptic curves and I believe also for abelian varieties). So any quotient which identifies complex conjugation and identity won't arise this way. $\endgroup$
    – Wojowu
    Apr 25, 2021 at 16:10
  • $\begingroup$ If your finite group has a 2-dimensional representation, then en.wikipedia.org/wiki/Serre%27s_modularity_conjecture implies that it comes from a modular form if it is odd etc. But only few of those come from elliptic curves. $\endgroup$ Apr 25, 2021 at 16:44
  • $\begingroup$ Even something stronger holds: The field obtained by adjoing all $E[p]$ to $\mathbb{Q}$, for all ellliptic curves $E$ and all $p$, is still very far away from the algebraic closure $\overline{\mathbb{Q}}$ (in fact it is a Hilbertian field). I know of no similar result when running over all abelian varieties though. $\endgroup$
    – Arno Fehm
    Apr 25, 2021 at 18:21

1 Answer 1

4
$\begingroup$

For elliptic curves, the answer is no, as Chris Wuthrich and Aron Fehm point out in the comments. In fact I think every extension with Galois group a finite simple group not of the form $PSL_2(\mathbb F_p)$ for any $p$ cannot arise this way.

For abelian varieties, the answer is positive. In fact it's sufficient to take just the $2$-torsion, and even just the $2$-torsion of Jacobians of hyperelliptic curves. Given an arbitrary extension, take the minimal polynomial $f$ of a generator $\alpha$ of that extension, form the hyperelliptic curve with equation $y^2 = f(x)$, and take its Jacobian. (If the degree of $f$ is at most $4$, you will need to add some linear factors as well to make it hyper-.) The Galois group of the $2$-torsion of the hyperilliptic curve will be a subgroup of $S_n$, generated by the Galois action on the roots of this polynomial (plus the point at $\infty$, if the degree is odd) and will therefore include the Galois group of your field.

$\endgroup$
2
  • $\begingroup$ What about e.g. abelian threefolds? Do you get all finite quotients? $\endgroup$
    – user178109
    Apr 26, 2021 at 5:41
  • $\begingroup$ @Oniqa: No bounded dimension will do. $\endgroup$
    – Arno Fehm
    Apr 26, 2021 at 6:55

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.