0

Say we have $C\subset \mathbb{CP}^2$, a smooth curve of degree 3. I am aware of the group structure on cubics, what I don't get, is why are all inflection points with only real coordinates lie on a single line in $\mathbb{CP}^2$.

I think I see that if two real inflection points are on a line, then the third point (which will also be an inflection point) will also only have real coordinates. And so if there are 4 real inflection points then all inflection points will be real.

  • There's no need to use abbreviations like "infpoints" here - you have time to write out all of your statements in detail :) 2. Do you know that an inflection point is a 3-torsion point for the group law? Do you know what the structure of $n$-torsion points on an elliptic curve is?
  • – KReiser Nov 16 '23 at 01:08
  • ok thanks :) $\$ 2. You mean that P+P+P=0 iff P is an inflection point? I don't learn maths in english, and I can't always find what the Hungarian version of the terminology is.
  • – Ilcu_elte_smurf Nov 16 '23 at 06:12
  • 1
    The answer here seems to be applicable. – KReiser Nov 17 '23 at 02:54
  • @KReiser Thanks for this reference. Another reference that I just discovered here , close to my approach. – Jean Marie Nov 17 '23 at 09:13
  • As @KReiser mentions my answer, I'll comment a few things here. Firstly, you have discovered a non-trivial fact about elliptic curves, the field generated by the $N$-torsion points must contain an $N^{\text{th}}$-root of unity. Since inflection points are $3$-torsion points on an elliptic curve (once one is chosen to be the identity) the field generated by all the inflection points must be $\mathbb{C}$ since primitive 3rd roots of unity are not rational. In general this is due to the existence of the Weil pairing, but for 3-torsion it is somewhat simpler as my answer indictates. – Mummy the turkey Nov 17 '23 at 11:03
  • Secondly, if you want to prove this yourself as an exercise, I would recommend starting with the Hessian description of the inflection points. First show that one inflection point must be over $\mathbb{R}$. Then move it to infinity to get a Weierstrass equation. Now compute the intersection with your cubic curve and project onto the $x$-coordinate (this will give you the $3$-division polynomial). Finally, compute its discriminant and see that its splitting field must contain a 3rd root of unity. I think this approach is quite illustrative and touches on some nice things. – Mummy the turkey Nov 17 '23 at 11:07