2
$\begingroup$

f:X-->Y is flat and projective map between integral varieties over k, an algebraically closed field. Suppose every fiber at closed points of Y is still an integral variety. L is a line bundle on X, if f_*(L) is trivial line bundle on Y, and for every closed point y of Y, L_y is also a trivial line bundle on fiber X_y, can we deduce that L is also a trivial line bundle on X?

$\endgroup$

1 Answer 1

6
$\begingroup$

Yes, you can. This follows from the see-saw principle for instance. You can also argue directly as follows. The spaces of global sections $H^{0}(Y,f_*L)$ and $H^{0}(X,L)$ are naturally isomorphic. Since $f_*L$ is a trivial line bundle we can choose a global nowhere vanishing section $e$ of $f_*L$. Let $s \in H^{0}(X,L)$ be the section of $L$ corresponding to $e$ under the above isomorphism. To show that $L$ is trivial it suffices to check that $s$ does not vanish anywhere. But if $x \in X$ is a closed point where $s$ vanishes, then if we restrict $s$ to the fiber $X_{f(x)}$, we will get a section of the structure sheaf of integral projective variety which vanishes at a point. Thus the restriction of $s$ to $X_{f(x)}$ must be identically zero. This shows that $e$ vanishes at $f(x)$ which is a contradiction.

$\endgroup$
1
  • $\begingroup$ Aha, surprised to see you here! Thanks a lot, I also thought about using this criterion for a line bundle's triviality, but I didn't realize that if a section for a trivial line bundle vanishes at one point, then it vanishes every where! I was misled by the flexible picture for real line bundles. $\endgroup$
    – TJCM
    Commented Dec 15, 2009 at 4:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .