MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This question is not very precise; I hope it is suitable for the site.

I have come to a situation where I have to study rational points on an elliptic curve defined over $\mathbb{Q}$. I don't know much about the curve, let alone its equation. I already have one rational point, which sits on a bounded real connected component. What I want to avoid is that this is the only rational point (other than the marked point).

I am not sure what to use about my curve that will help me get there, so I turn the questio the other way round:

What is known about elliptic curves $E$ over $\mathbb{Q}$ such that $E(\mathbb{Q}) \cong \mathbb{Z}/2 \mathbb{Z}$?

share|cite|improve this question
Torsion points are integral or half-integral in suitable natural coordinates (Lutz-Nagell thm. see, e.g., Silverman) so if you can prove that your point is not integral, then it's of infinite order. Similarly, if you can show it has non-zero canonical height. Your question is too vague. – Felipe Voloch May 13 '11 at 10:29
Yes, I know my question is vague. That is because I do not know much about this point. I was hoping that having Mordell-Weil group of order 2 would imply some restrictions, and then work backwards. – Andrea Ferretti May 13 '11 at 10:35
You will have to tell us more about your $E$ if you hope that we can help. Stated this way, your question is too vague. – Chris Wuthrich May 13 '11 at 11:08
The fact is, I don't know much more than stated in the question. I can probably extract more information, if I know what to look for. So, in a sense, any theorem starting like "Let $E/\mathbb{Q}$ be a an elliptic curve with $E(\mathbb{Q}) = \mathbb{Z}/(2)$; then..." would help me. – Andrea Ferretti May 13 '11 at 11:59
up vote 14 down vote accepted

Mazur's theorem ensures that there are exactly 15 possible cases for the torsion part of the Mordell-Weil group of an elliptic curve: the cyclic groups $\mathbb{Z}_n$ (with $1\leq n\leq 10$ or $n=12$) and the groups $\mathbb{Z}_2\times\mathbb{Z}_n$ for $n=2,4,6,8$.

In his paper Universal Bounds on The Torsion of Elliptic Curves, Proc. London. Math. Soc.(1976) 33, 193-237 , Daniel Sion Kubert (who was a student of Mazur) presents in table 3 (page 217) a list of parametrizations for the different possible cases.

In particular, curves with a $\mathbb{Z}_2$ torsion are parametrized by the following family: $$ \mathbb{Z}_2\ \text{torsion}:\quad y^2=x(x^2+a x+ b), \quad b^2(a^2-4b)\neq 0. $$

The example given in Francesco's answer is a special case with $a=0$. As another example, the case with torsion $\mathbb{Z}_2\times \mathbb{Z}_2$ is parametrized by the Legendre family:

$$ \mathbb{Z}_2\times\mathbb{Z}_2 \ \text{torsion}:\quad y^2=x(x+r)(x+s), \quad r\neq 0 \neq s \neq r. $$

A slight generalization of the Hesse family parametrizes the curves with torsion $\mathbb{Z}_3$:

$$ \mathbb{Z}_3 \ \text{torsion}:\quad y^2+a_1 x y +a_3 y =x^3, \quad a_3^3( a_1^3-27 a_3)\neq 0. $$ For the other groups you might have to use Tate's normal form $$ E(b,c): \quad y^2+(1-c)x y - b y =x^3- b x^2 $$ and the condition for a given torsion is expressed as an algebraic condition on $b$ and $c$.

For example for $\mathbb{Z}_4$, we have $c=0$ and $b^4(1+16b)\neq 0$, which gives: $$ \mathbb{Z}_4 \ \text{torsion}:\quad E(b,c=0): \quad y^2+x y - b y =x^3- b x^2, \quad b^4(1+16b)\neq 0. $$

For a review, you can read chapter 4 of the book of Husemoller . A friendly short review is also available in section 2 of this string theory paper by Aspinwall and Morrison ( they don't present all the 15 cases but for those they analyze, they express everything in Weierstrass form).

share|cite|improve this answer

I think not so much is known.

Conjecturally, elliptic curves (defined over $\mathbb{Q}$) of rank $0$ have density $1/2$, and since at any rate only a finite number of cases for the torsion part of $E(\mathbb{Q})$ are possible (by Mazur's celebrated result), I guess that elliptic curves with $E(\mathbb{Q}) \cong \mathbb{Z}/2 \mathbb{Z}$ should have density $1/2$ too.

For this reason, it seems to me that a complete classification is out of reach.

If you want to see an infinite family of cubic curves with this property, take

$y^2=x^3 + px$

where $p$ is a prime number such that $p \equiv 7$ (mod $16$), see [Silverman-Tate, Rational points on elliptic curves, p. 105].

share|cite|improve this answer
I don't think that the statement that curves with torsion groups of order 2 have density 1/2 is correct; it's probably, to paraphrase Pauli, not even false: these density statements only make sense if you think of elliptic curves as being ordered, and this can be done in several ways: with respect to the size of its coefficients, its discriminant, its conductor, etc. – Franz Lemmermeyer May 13 '11 at 10:38
In addition, torsion of order 2 means that the defining polynomial on the right hand side splits into a linear and a quadratic factor over the rationals, plus the rather complicated condition of the absence of further torsion points of odd order. – Franz Lemmermeyer May 13 '11 at 10:38
That's right. If we consider elliptic curves as polynomials $x^3+Ax+B$ with integral $A,B$ ordered by size, you should get density zero for those with roots. From this point of view (and I think most other reasonable ones), the density of elliptic curves with 2-torsion should be zero. – Minhyong Kim May 13 '11 at 10:45
Dear Franz and Minhyong, thank you for your comments. I'm actually not an expert in the field, so probably my idea of density is too vague and intuitive. What I really want to say in this answer in that elliptic curves with $E(\mathbb{Q})=\mathbb{Z}/2 \mathbb{Z}$ are probably "too many" to admith an exhaustive description. – Francesco Polizzi May 13 '11 at 11:00
@Francesco: it's not really a matter of size. Families of elliptic curves whose torsion parts have subgroups of certain orders have been classified. – Franz Lemmermeyer May 13 '11 at 17:51

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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