On the complement of nef divisors on projective manifolds
S. Feklistov
TL;DR
The paper characterizes when the complement of a nef effective divisor on a complex projective manifold fails the Hartogs extension property, proving this occurs precisely when the divisor is abundant with Iitaka dimension one. The authors analyze the problem via a case split on κ(D) and ν(D), combining Ueda theory, Lefschetz-type cohomology, and fibration arguments to derive a fiber-structure criterion: X is not Hartogs iff X admits a fibration over an affine curve. The results yield practical tests through linear systems and yield corollaries and examples on ruled surfaces, while suggesting pathways to extend the framework to Kähler manifolds and non-nef settings through divisorial decompositions. Overall, the work links geometric positivity, fibration theory, and complex-analytic extension phenomena to give a precise, transferable criterion for Hartogs in complements of nef divisors.
Abstract
Let $X'$ be a complex projective manifold, $\dim X'>1$, $Z$ a connected analytic subset of codimension one which is the support of a nef effective Cartier divisor $D$ on $X'$, $X:=X'\setminus Z$. Let $κ(D)$ be the Iitaka dimension of $D$. We prove that $X$ is not Hartogs if and only if $D$ is abundant and $κ(D)=1$. In particular, $X$ is not Hartogs if and only if $X$ is a proper fibration over an affine curve.
