Let $n\in \mathbb{Z}$ be not divisible by $15$ and suppose $n$ and $-n$ are both not perfect squares. Prove that at least one of the elliptic curves $Y^2 = X^3 + nX$ and $Y^2 = X^3 - nX$ has torsion group given by $\{{\bf o},(0,0)\}$ (where ${\bf o}$ is the point at infinity).

Here is my progress:

The discriminant is $\Delta = \pm 4n^3$ so we may use reduction with any odd prime $p \nmid n$. Since $n$ is not divisible by $15$, at least one of $3\nmid n$ and $5\nmid n$ holds.

  • Suppose $3\nmid n$ - then we can use $p=3$. Since $x^3 \equiv x \pmod 3$, for the reduced curves we can equivalently consider $Y^2 = (n+1)X$ and $Y^2 = (1-n)X$. If $n\equiv 1 \pmod 3$, then the former has only ${\bf {\underline{o}}}, (0,0), (2,\pm 1)$ as points, while the ones on the latter are ${\bf {\underline{o}}}, (0,0), (1,0), (2,0)$. The points are exchanged when $n\equiv 2 \pmod 3$.

  • Suppose $5\nmid n$ - then we can use $p=5$. For $n\equiv 1,2,3,4 \pmod 5$ a direct verification shows that $Y^2 = X^3 + nX$ has $4,2,6,8$ points respectively; hence $Y^2 = X^3 - nX$ has $8,6,2,4$ points, respectively.

Hence currently I have that at least one of the curves has $1, 2$ or $4$ torsion points. It is also easy to check that $(0,0)$ is always a point of order $2$. So how to rule out the case of $4$ torsion points?

Any help appreciated!

  • $\begingroup$ Both curves have only a single rational point of order two. So you only need to rule out the case of a cyclic torsion group of order four. I might try and check whether it is possible that $(0,0)=[2]P$ for a rational point $P$. In other words, can the tangent at a rational point pass through the origin? I don't know if this can be done though. $\endgroup$ Feb 1, 2020 at 14:44

1 Answer 1


Since $n$ and $-n$ are not perfect squares, the point $(0, 0)$ is the only point of order $2$ on both curves.

Now if $P = (x, y)$ is a rational point of order $4$ on the curve $Y^2 = X^3 - nX$, then by the doubling formula, we have $$X(2P) = \frac{x^4 + 2nx^2 + n^2}{4y^2}.$$ However, $2P$ must be the point $(0, 0)$, hence $x^4 + 2nx^2 + n^2 = (x^2 + n)^2$ must be zero, which means $-n = x^2$ is the square of a rational number, and therefore a perfect square. This contradicts the assumption.

Changing $n$ to $-n$ shows that the curve $Y^2 = X^3 + nX$ has no point of order $4$ either.


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.