1
$\begingroup$

Let $j:\mathbb{A}^{n}\rightarrow\mathbb{A}^{n}$ an open immersion over a field $k$. Is it an isomorphism?

$\endgroup$
1
  • 7
    $\begingroup$ Take a look at the the Ax-Grothendieck theorem! $\endgroup$
    – M P
    May 25, 2013 at 10:58

2 Answers 2

4
$\begingroup$

If $U \cong \mathbb A^n$ is the image of $j$, then the long exact sequence of a pair shows that $\mathbb A^n \setminus U$ has no cohomology in any degree, which is only possible if $U = \mathbb A^n$.

More generally (using heavier machinery), no variety $X$ is isomorphic to a Zariski open subset $U \subsetneq X$. The same long exact sequence shows that the Hodge-Deligne polynomial of $X \setminus U$ would need to vanish, which is only possible for $X \setminus U = \varnothing$.

$\endgroup$
1
  • 4
    $\begingroup$ For the second part, it is easier to spread on a finitely generated domain, then count points in fibers. $\endgroup$
    – Angelo
    May 25, 2013 at 15:18
14
$\begingroup$

Let $U \cong \mathbb{A}^n$ be the image of $j$ and let $\Delta = \mathbb{A}^n \setminus U$. Since $U$ is affine, $\Delta$ is purely codimension $1$, so $\Delta$ is the zero locus of some $f \in k[x_1, \ldots, x_n]$. But then $f$ is a unit on $U$, contradicting that $U \cong \mathbb{A}^n$.

$\endgroup$

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.