1
$\begingroup$

Assume we have a smooth projective variety $X$ over a field (finite field if matters). Consider a smooth hypersurface section $H$ of $X$. Let $Spec(A)=U=X \setminus H$ be an affine variety. I want to know whether there are any nice description of vector bundles on $X$ such that they are trivial on $U$ and $H$? My suspicion was that they are all direct sums of line bundles of the form $O(nH)$ is that correct?

Edit: Let's assume that we are talking about rank $m$ vector bundles that are subsheaves of the constant sheaf given by the vector space $F(X)^{\oplus m}$ which $F(X)$ is the function field or equivalently field of fractions of $A$. Does that change anything?

$\endgroup$
5
  • 2
    $\begingroup$ This is not true. Try the projective plane and $H$ a hyeprplane. $\endgroup$ – Mohan Apr 10 '19 at 1:33
  • $\begingroup$ @Mohan Is this true if we assume the variety is 1 dimensional? And assume the $H$ is just 1 point. $\endgroup$ – user127776 Apr 10 '19 at 3:29
  • 5
    $\begingroup$ This is definitely false. Take a smooth projective curve $X$ of genus $\geq 2$. There are of course many nontrivial vector bundles on $X$ with trivial determinant. Their restriction to the complement of a point is trivial, because a projective module over a Dedekind ring with trivial determinant is free. $\endgroup$ – abx Apr 10 '19 at 5:24
  • $\begingroup$ I made an edit, I'd really appreciate your opinions on it. $\endgroup$ – user127776 Apr 10 '19 at 6:55
  • $\begingroup$ All rank $m$ vector bundles embed into $F(X)^{\oplus m}$ (by restriction to the generic fiber). $\endgroup$ – abx Apr 10 '19 at 7:07
1
$\begingroup$

In many cases, $\mathcal{O}_X(n \cdot H)$ itself will not be trivial on both $H$ and $U$, because it will be nontrivial on $H$. For example, if $H \subset \mathbb{P}^n$ is a degree $d$ hypersurface, then $ \mathcal{O}_{\mathbb{P}^n}(H) \simeq \mathcal{O}_{\mathbb{P}^n}(d)$ and so $\mathcal{O}_{\mathbb{P}^n}(H)|_H \simeq \mathcal{O}_H(d)$ which is non-trivial as long as $n> 1$, for instance because it is ample.

EDIT: I realize what this shows is: not all sums of line bundles of the form $\cal{O}_X(n \cdot H)$ are trivial on both $U$ and $H$. All the same it seemed relevant.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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