I've been playing around with Riemann surfaces of cubics, and it seems to me that all surfaces obtained as coverings of the Riemann sphere from equations of the form $w^3 = q(z)$, where $q(z)$ is a cubic with three distinct roots, must be isomorphic.

Argument: we have a critical point of multiplicity 3 at each of the roots of $q(z)$. Monodromy around a small counterclockwise circuit about any of these points multiplies $w$ by the same cube root of unity ($\neq 1$). So monodromy around a circuit enclosing all three roots of $q(z)$ leaves $w$ unchanged, so no branch points over $\infty$.

Now we can move the three roots of $q(z)$ to any other positions using a Möbius transformation, so by the previous paragraph, we should be able to establish an analytic isomorphism between the surfaces via continuation.

Is this correct? If so, what is the common J-invariant? And since we can choose $q(z)=z^3-1$, this means that all these curves have CM, right?

References appreciated.

  • 1
    $\begingroup$ Your argument seems good, but it's not clear to me how to change $w^3 = z^3 - 1$ into $w^3 = 2(z^3 - 1)$ (i.e., I'm suspicious that your proof really applies only to monic polynomials $q(z)$). Perhaps I haven't thought about it long enough. $\endgroup$ – John Hughes Dec 23 '13 at 23:30
  • $\begingroup$ @John: Good point. However, I think rescaling will handle this. Just replace $w$ with $rw$ and divide the whole equation by $r^3$. Now let $r$ be the leading coefficient of $q(z)$. $\endgroup$ – Michael Weiss Dec 24 '13 at 0:28

This is a great observation!

Let $E/\mathbb{Q}$ be a curve given by a model $v^3=q(u)$, for some cubic polynomial $q\in\mathbb{Q}[u]$, and assume that the projective closure of this model, i.e., $V^3=W^3q(U/W)$, is smooth, and it has a rational point $P$. Then, $(E,P)$ is an elliptic curve defined over $\mathbb{Q}$. Moreover, $E$ admits an endomorphism $$[\rho] : E\to E$$ that sends $[U,V,W]$ to $[U,\rho V,W]$, where $\rho$ is a primitive third root of unity. Hence, $\operatorname{End}(E)$ is strictly larger than $\mathbb{Z}$, and so $E/\mathbb{Q}$ is a curve with complex multiplication. Further, $$\operatorname{End}(E)\otimes \mathbb{Q} \cong \mathbb{Q}(\rho),$$ and therefore the complex multiplication is by an order in the ring of integers of the quadratic imaginary field $\mathbb{Q}(\rho)=\mathbb{Q}(\sqrt{-3})$. It is a fact that follows from the theory of complex multiplication that all the elliptic curves with complex multiplication by $\mathbb{Q}(\sqrt{-3})$ have $j$-invariant equal to $0$. It follows that all the elliptic curves of the form $V^3=W^3q(U/W)$ are isomorphic over $\mathbb{C}$ (as the $j$-invariant classifies elliptic curves up to isomorphism over $\mathbb{C}$).

For example, consider $E: V^3=U^3-W^3$ (or $v^3=u^3-1$ in affine coordinates), with $P=[1,1,0]$. Then, $E$ is isomorphic to $$E': Y^2Z-9YZ^2=X^3-27Z^3$$ or $y^2 - 9y = x^3 - 27$ in affine coordinates, via the map $\phi:E'\to E$ that sends $$\phi([X,Y,Z])=(Y-9Z,Y,3X),$$ which sends $[0,1,0]$ to $[1,1,0]$. The curve $E'$ is now easily checked to have $j=0$ via the usual formulas, and therefore it has complex multiplication by $\mathbb{Q}(\sqrt{-3})$.

  • $\begingroup$ Thanks! What reference would you recommend for the result about CM by $\rho$ implying $j=0$? $\endgroup$ – Michael Weiss Feb 26 '14 at 18:42
  • $\begingroup$ See Silverman's "Advanced Topics in the Arithmetic of Elliptic Curves", Chapter 2, Example 1.3.2. $\endgroup$ – Álvaro Lozano-Robledo Feb 26 '14 at 19:02

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.