1
$\begingroup$

M. Reid has a theorem as following:

Let $X$ be a non-singular projective complex surface, $L \subset \Omega^1_X$ be a line bundle. If $h^0(L^{\otimes n})\geq2$ for some $n\geq 1$, then there is a moephism $f:X\rightarrow C$ of $X$ onto a nonsingular curve $C$ of positive genus, and a ramification divisor $D$ (of the map $f$) such that $L$ and $f^*(\Omega^1_C)(D)$ coincide as sheaves.

How to derive the following statements?

(a) there exists a constant $k$ such that for all $n$ $$h^0(L^{\otimes n} )< nk$$ (b) $L$ is not in the positive cone of $NS(X)$

$\endgroup$

2 Answers 2

1
$\begingroup$

(a) The ramification divisor $D$ is contained in some fibers of $f$, so we can write $L=f^*M(-E)$, with $M$ a line bundle on $C$ and $E$ effective. Therefore $h^0(L^{{\scriptscriptstyle\otimes} n })$ is bounded by $h^0(f^*M^{{\scriptscriptstyle\otimes} n })=h^0(M^{{\scriptscriptstyle\otimes} n })$ (projection formula).

(b) What do you call the positive cone? We have $(L)^2\leq 0$ and $(L\cdot F)=0$ for $F$ a general fiber of $f$, so certainly $L$ is not ample. On the other hand if $D=0$ $L$ is nef.

$\endgroup$
1
  • $\begingroup$ Thank you very much.But why can we write $L$ as $f^*M(-E)$ with $E$ effective. By condition $L=f^*(\Omega_C^1)(D)$, what is the relation between $M$ and $\Omega_C^1$. We know that on a projective surface $X, \mathcal{C}=\{\omega \in H^{1,1}(X,\mathbb{R})|\omega^2>0\}$ consists of two disjoint connnected cones. The positive cone is the component of $\mathcal{C}$ containing the class of kahler form. Can you Explain (b) in more detail, thank you. $\endgroup$
    – swalker
    Oct 15, 2013 at 10:30
2
$\begingroup$

I'm not entirely sure what you mean by a ramification divisor when a surface maps to a curve....

On the other hand, by the Bogomolov-Sommese vanishing theorem if $L\subset\Omega_X^1$, then the Kodaira dimension of $L$ is at most $1$. This implies both statements.


Remark: If by the ramification divisor you mean the union of the singular fibers, then the statement of the result you are citing implies that $L$ is contained in the pull-back of a line bundle on the curve and this implies the bound on the Kodaira dimension, so you don't need to use the Bogomolov-Sommese theorem. Part (a) is essentially that statement that the Kodaira dimension of $L$ is at most $1$ and that implies that it cannot be ample. Of course, it could still be a limit of amples, so it is not in the positive cone, but it may be on its boundary.

$\endgroup$

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.