0
$\begingroup$

Let $X$ be a smooth projective variety, $V, W$ closed subschemes in $X$ such that $V \cap W$ is finitely many points. Let $\mathcal{L}$ be a line bundle on $X$. Is there any relation between $h^0(\mathcal{L} \otimes_{\mathcal{O}_X} \mathcal{O}_{V.W})$ and the intersection multiplicity of V.W (like is the former bounded by the later)? Is so, can we say anything similar if $\mathcal{L}$ is a locally free sheaf not necessarily of rank $1$?

$\endgroup$
3
  • $\begingroup$ What is $\mathcal{O}_{V.W}$? $\endgroup$
    – Sasha
    Feb 23, 2015 at 18:55
  • $\begingroup$ @Sasha: It is the cokernel of the natural morphism $\mathcal{I}_V+\mathcal{I}_W \to \mathcal{O}_X$, where $\mathcal{I}_V, \mathcal{I}_W$ are the ideal sheaves of $V$ and $W$ respectively. Another definition, that one can use is $\mathcal{O}_V \otimes_{\mathcal{O}_X} \mathcal{O}_W$. $\endgroup$
    – Kali
    Feb 23, 2015 at 19:39
  • $\begingroup$ So, this is just the structure sheaf of the scheme-theoretical intersection. $\endgroup$
    – Sasha
    Feb 23, 2015 at 20:46

1 Answer 1

1
$\begingroup$

If $V$ and $W$ are Cohen--Macaulay, then $V \cap W$ is a finite number of points implies $Tor_{>0}(O_V,O_W) = 0$. This means that $V\cdot W = \ell(O_V \otimes O_W) = \ell(L \otimes O_V \otimes O_W)$ which is equal to the $h^0$ you are interested in.

$\endgroup$
2
  • $\begingroup$ Thank you very much. This is exactly the setting I am interested in. $\endgroup$
    – Kali
    Feb 23, 2015 at 22:10
  • $\begingroup$ Non-example: Let $V$ be a union of two planes at a point inside $\mathbb P^4$ (so not C-M), and $W$ a plane going through that point. Then the intersection multiplicity is of course $2$, but the scheme-theoretic intersection is a fat point of length $3$. To see the problem, let $W = W_1 \cap W_2$ be an intersection of hyperplanes. Then $V\cap W_1$ is a union of two lines and an embedded point, and $(V\cap W_1)\cap W_2$ has a point for each line (that's good) and also the embedded point (that's bad, and in particular, not transverse). $\endgroup$ Feb 24, 2015 at 5:35

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.