Local Positivity of Ample Line Bundles

TL;DR

This paper establishes a uniform lower bound of 1/n for the Seshadri constant of ample line bundles at very general points on smooth projective varieties, linking local positivity to global geometric properties.

Contribution

It proves a surprising uniform lower bound for Seshadri constants at very general points, advancing understanding of local positivity of ample line bundles.

Findings

Lower bound of 1/n for Seshadri constants at very general points

Existence of countably many subvarieties where the bound may not hold

Applications to adjoint and pluricanonical linear series

Abstract

Let $L$ be a nef line bundle on a smooth complex projective variety $X$ of dimension $n$. Demailly has introduced a very interesting invariant --- the Seshadri constant $\epsilon(L,x)$ --- which in effect measures how positive $L$ is locally near a given point $x \in X$. For instance, Seshadri's criterion for ampleness may be phrased as stating that $L$ is ample if and only if there exists a positive number $e > 0$ such that $\epsilon(L,x) > e$ for all $x \in X$, and if $L$ is VERY ample, then $\epsilon(L,x) \ge 1$ for every $x$. We prove the somewhat surprising result that in each dimension $n$ there is a uniform lower bound on the Seshadri constant of an ample line bundle $L$ at a very general point of $X$. Specifically, $\epsilon(L,x) \ge (1/n) $ for all $x \in X$ outside the union of countably many proper subvarieties of $X$. Examples of Miranda show that there cannot exist a bound (independent of $X$ and $L$) that holds at every point. The proof draws inspiration from two sources: first, the arguments used to prove boundedness of Fano manifolds of Picard number one; and secondly some of the geometric ideas involving zero-estimates appearing in the work of Faltings and others on Diophantine approximation and transcendence theory. We give some elementary applications of the main theorem to adjoint and pluricanonical linear series.