13
$\begingroup$

I asked this question on MSE but didn't get any response, so I'm asking here. I apologize in advance if this question is not research level.

A Fermat Curve of degree $n$ is the set of solutions to $x^n+y^n=z^n$, $x,y,z\in \mathbb R$. In this question, the OP provides a substitution which relates a Fermat Curve of degree $n=3,4$ to two different elliptic curves. To transform the Fermat Curve of degree $3$, the substitutions $$ a=\frac{12z}{x+y},\quad b=\frac{36(x-y)}{x+y} $$ produce $b^2=a^3-432$, an elliptic curve. Similarly for the Fermat Curve of degree $4$, the substitutions $$ a=\frac{2(y^2+z^2)}{x^2},\quad b=\frac{4y(y^2+z^2)}{x^3} $$ give $b^2=a^3-4a$. However, the substitutions used are not at all obvious, which leads me to wonder,

Is there a similar substitution which can relate a Fermat curve of arbitrary degree to an elliptic curve?

This is equivalent to asking whether there is always a nonconstant morphism from a Fermat curve to an elliptic curve.

Thank you in advance!

$\endgroup$
  • 6
    $\begingroup$ @AdamP.Goucher The OP is asking for a morphism from the Fermat curve onto an elliptic curve, not an isomorphism. Equivalently, does the Jacobian of the Fermat curve have an elliptic factor? $\endgroup$ – Joe Silverman Dec 20 '18 at 2:54
  • 7
    $\begingroup$ The cubic Fermat curve is a smooth cubic in $\mathbb P^2$, so has genus 1. It has the rational point $[1,-1,0]$. Hence it is isomorphic to an elliptic curve given by a Weierstrass equation. Finding the transformation is standard. The quartic Fermat curve maps 2-to-1 to the curve $C:u^4+1=v^2$ via $u=x/y$ and $v=z^2/y$. The curve $C$ also has genus 1 and a rational point $(u,v)=(0,1)$, hence it too can be mapped to an elliptic curve in Weierstrass form (using 19th century formulas!) . For higher $n$, there are lower genus curves that the Fermat curve maps to, but they're generally not elliptic. $\endgroup$ – Joe Silverman Dec 20 '18 at 3:00
  • 5
    $\begingroup$ The Jacobian of the Fermat curve decomposes (up to isogeny) as a product of CM abelian varieties. Accordingly the motive of the Fermat curve $X_N$ decomposes as a direct sum of motives associated to Hecke characters of the cyclotomic field $\mathbf{Q}(\zeta_N)$ (see Otsubo's work e.g. On special values of Jacobi-sum Hecke $L$-functions). So your question is about rationality properties of these Hecke characters. $\endgroup$ – François Brunault Dec 20 '18 at 9:24
  • 6
    $\begingroup$ I don't know the answer to the question but here is a reference: Koblitz, Rohrlich, Simple factors in the Jacobian of a Fermat curve. If I understand Theorem 2 correctly, then if $N \geq 5$ is a prime $\equiv 2 \textrm{ mod } 3$ then the simple factors of the Jacobian of $X_N$ all have dimension $(N-1)/2$. So for exemple $X_5$ and $X_{11}$ do not map to any elliptic curve. $\endgroup$ – François Brunault Dec 20 '18 at 9:38
  • 5
    $\begingroup$ @FrançoisBrunault: I think you should write your comment as an answer, this is actually quite nontrivial. The paper you quote is freely available at https://cms.math.ca/openaccess/cjm/v30/cjm1978v30.1183-1205.pdf. $\endgroup$ – abx Dec 20 '18 at 10:08
19
$\begingroup$

The following reference seems to answer the question of which Fermat curves admit a non-constant map to an elliptic curve:

Neal Koblitz, David Rohrlich, Simple factors in the Jacobian of a Fermat curve, Canadian Journal of Mathematics 30 No. 6 (1978) pp. 1183–1205, doi:10.4153/CJM-1978-099-6 (free pdf).

As a particular case of Theorem 2 there, if $N \geq 5$ is a prime $\equiv 2 \textrm{ mod } 3$ then the simple factors of the Jacobian $J_N$ of the Fermat curve $X_N$ all have dimension $(N−1)/2$. So for example, the Fermat curves $X_5$ and $X_{11}$ do not map to any elliptic curve.

The motive of the Fermat curve $X_N$ has been extensively studied: it decomposes as a direct sum of motives associated to Hecke characters of the cyclotomic field $\mathbf{Q}(\zeta_N)$, see e.g. Otsubo's work On special values of Jacobi-sum Hecke $L$-functions. Note also that every elliptic factor of a Fermat curve must have complex multiplication, essentially because $X_N$, and thus its Jacobian $J_N$, admits an action of the rather large group $\mu_N \times \mu_N$, where $\mu_N$ is the group of $N$-th roots of unity in $\mathbf{C}$. More generally, the factors of $J_N$ are CM abelian varieties.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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