Given an integer $d\geq 2$ is there an algebraic curve $C/\mathbb{Q}$ with $C(\mathbb{Q})\neq\emptyset$ and the natural map $C(\mathbb{Q})\to C(F)$ bijective for all number fields of degree at most $d$?
2 Answers
Here's a family of examples that are geometrically irreducible. Let $p$ be a prime number and consider the modular curve $X_{1}(p)$. If $F$ is a number field, the points in $X_{1}(p)(F)$ are either non-cuspidal (and in bijection with elliptic curves $E/F$ that have a $F$-rational point of order $p$), or cuspidal. There are $p-1$ cusps of which $\frac{p-1}{2}$ are rational, and the other $\frac{p-1}{2}$ are in a single Galois orbit. (They are rational over $\mathbb{Q}(\zeta_{p} + \zeta_{p}^{-1})$.)
Merel's proof of the uniform boundedness conjecture implies that if $d$ is fixed, then there is a constant $C = C(d)$ so that $|E(F)_{{\rm tors}}| \leq C$ for all number fields $F$ of degree $d$. If we choose a prime $p > \max \{2d+1, C \}$, then $X_{1}(p)(F)$ will have no noncuspidal points for any number field $F$ with $[F : \mathbb{Q}] \leq d$. Also, $X_{1}(p)(F)$ will contain no non-rational cuspidal points because these are defined over a number field of degree $\frac{p-1}{2} > d$.
In the case of affine algebraic curves, homogenizing a polynomial with no roots in a number field of degree up to $d$ should do. For example, $x^3 + y^2x + y^3 = 0$ has a rational point $(0, 0)$ and no other points in any quadratic number field.
-
4$\begingroup$ Is it geometrically irreducible? $\endgroup$ Commented Sep 21, 2021 at 7:22
-
$\begingroup$ These examples aren't geometrically irreducible, since they factor over the splitting field of the dehomogenized polynomial. The curves in Jeremy Rouse's answer are not only geometrically irreducible, but also smooth projective curves. $\endgroup$ Commented Sep 24, 2021 at 4:29
-
$\begingroup$ @JohnDoyle a much better answer indeed! $\endgroup$– AstroNiCommented Sep 27, 2021 at 19:15