A Quick Review:

The complex elliptic curve $\mathbb{C}/(\mathbb{Z} + \tau \mathbb{Z})$ may be rewritten using the exponential, $\text{exp(}{2 \pi i \tau}) =: q$ as $\mathbb{C}^\times/q^\mathbb{Z}$ .

The Tate-Jacobi theorem states:

There is a canonical formal group law associated to the complex elliptic curve $C^\times/q^{\mathbb{Z}}$ whose coefficients lie in the subring $\mathbb{Z}[E_4(q), E_6(q)]$ of $\mathbb{C}$, where the Eisenstein series $E_4$ and $E_6$ are convergent power series in $q$. After adjoining $q$, this formal group law becomes canonically isomorphic to the multiplicative one.

If I understand correctly, this theorem allows us to, among other things, continuously deform the multiplicative formal group law into a height 1 elliptic curve formal group law.

If we are to view the curve $\mathbb{C}^\times/q^\mathbb{Z}$ as a continuous family of groups fibered over the interval $[0,1)$

  • at $q=0$, we have the punctured complex plane, $\mathbb{C}^\times$, i.e., $\mathbb{G}_m$
  • for each $q$ such that $0 < |q| < 1$, we have ordinary elliptic curves

This seems analogous to the deformation of a supersingular elliptic curve to an ordinary one:

$$y^2 = 4x^3 + tx^2 + 2x$$

  • at $t = 0$ we have a supersingular elliptic curve, $y^2 = x^3 + x$ (up to rescaling of $x$) over Spec $W(F_9)[t]$
  • at $t \neq 0$, $t \in$ Spec $W(F_9)[t]$, we have ordinary elliptic curves

Here's my question: Do we have a "Tate-Jacobi-like" theorem for such a height 2 to height 1 family of elliptic curves (for some subring of a finite field, rather than a subring of $\mathbb{C}$)?

That is, how can we continuously deform a height 2 formal group law into a height 1 formal group law (in a somewhat canonical way)?

  • $\begingroup$ Hi Catherine, what do you mean by $y^2=x^3+x$ being supersingular? meaning, at what prime? $\endgroup$ Sep 17 '15 at 17:16
  • $\begingroup$ Yes, $y^2=x^3-x$ is supersingular exactly for the primes $\equiv3\pmod4$. $\endgroup$
    – Lubin
    Sep 17 '15 at 18:58
  • $\begingroup$ Hi Bruno, my mistake. I meant for the coefficient ring to be the Witt ring of $F_9$, which I've been told to think of like $Z_p[x]$, where $Z_p$ is the padics. $\endgroup$ Sep 17 '15 at 19:01
  • 1
    $\begingroup$ Indeed, $W(\Bbb F_9)$ is just the quadratic unramified over $\Bbb Z_3$. No “$x$”. $\endgroup$
    – Lubin
    Sep 17 '15 at 19:05

Here’s a worm’s-eye view, perhaps not much help to you. You may always start with a supersingular curve in characteristic $p$, like $p=13$, where $y^2=x^3+x+4$ is supersingular. I choose $13$, because neither $y^2=x^3-x$ nor $y^2=x^3+1$ is supersingular there. Then you may always take your constant supersingular curve and jiggle it algebraically, here you could take $y^2=x^3+(1+t)x+4$ over $\Bbb F_{13}[t]$, to get a curve that’s ordinary for all but a finite number of values of $t$, among them $t=0$. Then, if you wish, you can take $y^2=x^3+(1+t)x+4$ as a curve over $W(\Bbb F_{13})[t]$. But I think you probably want something less ad hoc.


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.