8
$\begingroup$

It is not hard to construct such curves explicitly, e.g. my favourite example is a curve $U$ singular at $(3:4:5i)$ and also passing through $(1:0:0)$, $(0:1:0)$, $(0:0:1)$, $(1:1:0)$, $(1:0:1)$, $(0:1:1)$. (Amusingly, one more real point on $U$ is also rational, namely $(3^4:4^4:5^4)$; two more are quadratic irrational).

Is there an easier argument demonstrating that such curves exist? A reference would be good enough, too.


Edit: It is immediate from Bezout theorem that a curve $W$ with more than 9 real points must contain a real oval (i.e. a curve in $\mathbb{RP^2}$), just look at the intersection of $W$ and its complex conjugate $W^*$. A similar argument (see Noam's answer) shows that if it has 8 real points it must also have 9 of them (counting multiplicities, as usual).

$\endgroup$

1 Answer 1

14
$\begingroup$

Let $P_1,\ldots,P_8$ be "random" real points (which could even be rational). Then the space of cubic polynomials vanishing on all $P_i$ has dimension $10 - 8 = 2$. Let $(C_1, C_2)$ be a basis of this real vector space. Then the cubic curves $C_1=0$ and $C_2=0$ meet at $P_1,\ldots,P_8$ and at some ninth point $P_9$, which is automatically real (and also rational if $P_1,\ldots,P_8$ were). The cubic $C_1 + i C_2 = 0$ then passes through $P_1,\ldots,P_9$ and has no other real point.

$\endgroup$
2
  • $\begingroup$ Well, this is the intuition (one can probably even replace "random" by "no 4 on a line, no 7 on a conic"). But I don't know how one does make this into a proof. $\endgroup$ Jun 13, 2018 at 16:37
  • $\begingroup$ OK --- indeed, all you need is that the space of real cubics vanishing on the 8 points is 2-dimensional, and $C_1$ and $C_2$ do not share a component. $\endgroup$ Jun 13, 2018 at 17:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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