6
$\begingroup$

Let $k$ be an algebraically closed field. Let $H_1, H_2$ be two smooth hypersurfaces of the same degree $d$ in $P^n_k$. Let $U_1,U_2$ be their complements respectively. Are $U_1,U_2$ isomorphic as algebraic varieties?

In $n=1,d=1$ case this is true, because the complement of any point is isomorphic to $A^1$.

But $n=2$ case I guess this might be false. I want to prove that if $U_1,U_2$ are isomorphic then they must be induced by an automorphism of $P^n$, but this seems hard. I read something about the 'complement problem' on enter link description here, but this seems to be a more complicated question, and it focuses on $A^n$ instead.

Maybe the $n=2,d=3$ case is easier? In this case, elliptic curves are isomorphic if and only if they have the same $j$-invariant. Can we read this from its complement?

Are there any solutions/counterexamples? Any comments are welcome!

$\endgroup$
0

3 Answers 3

9
$\begingroup$

The answer is no. Perhaps the simplest case is $n=2$, $d=4$. There is a unique double covering $\pi _i:S_i\rightarrow \mathbb{P}^2$ branched along $H_i$. If $U_1$ and $U_2$ are isomorphic, $S_1$ and $S_2$ are isomorphic; then $H_1$ and $H_2$ are isomorphic, because $H_i$ is the branch locus of the morphism $\pi _i$, which is given by the anticanonical system.

$\endgroup$
2
  • $\begingroup$ Why must $S_1$ and $S_2$ be isomorphic if $U_1$ and $U_2$ are isomorphic? $\endgroup$
    – Yuan Yang
    Jul 4, 2021 at 21:04
  • $\begingroup$ Let $\tilde{U}_i:=\pi _i^{-1}(U_i) $. Since $\pi _1(U_i)=\mathbb{Z}/4$, the étale double covering $\tilde{U}_i\rightarrow U_i$ is canonical, so $\tilde{U}_1\cong \tilde{U}_2$. Then $S_1\cong S_2$ because $S_i$ is the normalization of $\mathbb{P}^2$ in $\tilde{U} _i$ (stacks project, 29.53). $\endgroup$
    – abx
    Jul 5, 2021 at 3:49
10
$\begingroup$

If $U_1$ and $U_2$ are isomorphic then $H_1$ and $H_2$ are equal in the Grothendieck ring of varieties and thus, by the Larsen-Lunts theorem, stably birational, which if $d>n$ implies that they are isomorphic.

This is probably extreme overkill, but it seems to handle some different cases than the other answers.

$\endgroup$
1
  • $\begingroup$ This is incisive and elegant! I don't even know if Mr.Yoshihara noticed this when he wrote his papers. He dealt with some more non-smooth cases though. $\endgroup$
    – Yuan Yang
    Jul 4, 2021 at 20:44
7
$\begingroup$

The easiest case is $n = 1$, $d = 4$. Indeed, the embeddings $U_i \to \mathbb{P}^1$ are canonical, hence an isomorphism $U_1 \cong U_2$ extends to an isomorphism of the ambient projective lines and induces an isomorphism $$ \mathbb{P}^1 \setminus U_1 \cong \mathbb{P}^1 \setminus U_2. $$ So, if the cross-ratio of the four points $\mathbb{P}^1 \setminus U_1$ differs from the cross-ration of $\mathbb{P}^1 \setminus U_2$, there can't be such an isomorphism.

$\endgroup$
1
  • $\begingroup$ Indeed. If we allow these hypersurfaces to be reducible it seems their complements are less likely to be isomorphic. $\endgroup$
    – Yuan Yang
    Jul 4, 2021 at 20:49

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.