Finding torsion subgroups of elliptic curves over finite fields.

Given $y^2=x^3+x+1$ over $F_3$ I need torsion subgroup of $E[3]$

$E[3]$ is either trivial or isomorphic to $\mathbb Z_3$

The points $(1,0),(-1,0),(0,0)$ are each of order $2$, so useless, but the point $(3,1)$ has order $4$ i.e. $4(3,1)=(\infty,\infty)$, Is there a possibility to combine this point with the others (or with itself, I cannot see it now) to get an element of order $3$ ?


To answer your specific question, no, you cannot "combine" points of order 2 and order 4 to create points of order 3. On the other hand, it's rather easy to find the points of order 3. Simply use the duplication formula to write $$x(2P)=x(P).$$ Clearing denominators will give you an equation to solve for $x(P)$. (In general, you'd get a quartic equation, but since you're looking for $p$-torsion in characteristic $p$, the degree will be lower.) If you get a constant, the curve is supersingular, and $E[3]=0$. If you get a non-constant equation, then $E[3]\cong\mathbb{Z}/3\mathbb{Z}$.

  • $\begingroup$ But this curve is supersingular, so the $3$-torsion is trivial. And generally, the $3$-torsion points are the points of inflection, easy to find. $\endgroup$ – Lubin Nov 12 '16 at 12:21
  • $\begingroup$ @Lubin True, but I was trying to explain how one would solve such problems in general, not specific to this particular curve and prime. And even for 3-torsion, if one knows the duplication formula, it's certainly no harder to set $x(2P)=x(P)$ than it is to rederive the equation for a point to be an inflection point. I think it was John T who showed me the trick of finding an (admittedly singular) equation for $X_1(2n+1)$ by setting $x((n+1)P)=x(nP)$. $\endgroup$ – Joe Silverman Nov 12 '16 at 13:12

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.