2
$\begingroup$

Let $f:X\dashrightarrow C$ be a rational fibration from a 3-dimensional variety $X$ to a curve $C$ such that generic fiber is smooth and different fibers intrsect in smooth curves. Take $S$ to be a smooth fiber. Let $D$ be the sum of intersections of $S$ with all other fibers. It is an effective divisor on $S$.

Is it true that $N_{S/X}\cong \mathcal{O}_S(D)$?

If this is not true, maybe something weaker is true. For instance, is $H^0(N_{S/X})$ is always nonzero? Maybe it is even true that locally fibers are deformations of each other? Is it at least true that $c_1(N_{S/X})=D$?

Any other information about this normal bundle is also very interesting.

Edit(thanks to @potentially dense): Probably this was unclear in the first version of the question. By fiber of a rational morphism I mean the closure of a fiber of the regular morphism $f:U\to C$ where $U$ is the domain of definition of $f$.

$\endgroup$
3
  • 1
    $\begingroup$ How can two different fibers intersect? $\endgroup$
    – Henri
    May 20, 2016 at 0:43
  • $\begingroup$ One needs to be clear exactly what "fibre" means in this case. I guess the OP is using it to mean the closure in $X$ of a fibre of the restricted morphism $f_{|U} : U \rightarrow C$ where $U$ is the maximal open set on which $f$ is defined. (In which case, fibres can certainly intersect.) $\endgroup$ May 20, 2016 at 8:47
  • $\begingroup$ @potentiallydense Sure, this is exactly what I meant, thank you. $\endgroup$
    – lks8271
    May 20, 2016 at 8:56

1 Answer 1

3
$\begingroup$

If $C$ is not rational and $X$ is smooth, the map $f$ is a morphism. Indeed, you can view $f$ as a map into the Jacobian $J(C)$ and a map from a smooth variety to an abelian variety is always a morphism.

So if $C$ is not rational, the normal bundle to a fiber is trivial.

If $C={\mathbb P^1}$ then $f$ is given by a $1$-dimensional linear system $|S|$ and the normal bundle to an element $S\in |S|$ is ${\mathcal O}_S(S)$.

(I am assuming that $f$ has connected fibers; you can always reduce to this case by considering the Stein factorization)

$\endgroup$

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.