Constructing new ample divisors out of old ones

DOI10.1215/S0012-7094-99-09803-4MaRDI QIDQ1974922zbMATH OpenFDO

Publication date 27 March 2000

Published in Duke Mathematical Journal (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/alg-geom/9705018

algorithm ample cone ample divisor blowing-up Nagata's conjecture Picard group rational surface rational surfaces remainder symplectic manifold

Mathematics Subject Classification ID

Divisors, linear systems, invertible sheaves (14C20) Picard groups (14C22) Birational automorphisms, Cremona group and generalizations (14E07) Rational and ruled surfaces (14J26) Computational aspects of higher-dimensional varieties (14Q15)

Abstract: We prove a gluing theorem which allows to construct an ample divisor on a rational surface from two given ample divisors on simpler surfaces. This theorem combined with the Cremona action on the ample cone gives rise to an algorithm for constructing new ample divisors. We then propose a conjecture relating continued fractions approximations and Seshadri-like constants of line bundles over rational surfaces. By applying our algorithm recursively we verify our conjecture in many cases and obtain new asymptotic estimates on these constants. Finally, we explain the intuition behind the gluing theorem in terms of symplectic geometry and propose generalizations.

Review text

The author studies the ample cone of surfaces \(S\) obtained by blowing-up \(\mathbb{P}^2\) at \(k\) distinct points \(p_1,\dots,p_k\). A vector \((d;m_1, \dots, m_k)\in \mathbb{Z}_+ \times \mathbb{Z}^k_{\geq 0}\) is said to be ample (respectively nef) if there exists a surface \(S\) as above such that \(\pi^*{\mathcal O}_{\mathbb{P}^2} (d)-\sum^k_{j=1} m_jE_j\) is ample (respectively nef), where \(\pi\) is the blow-up and \(E_j=\pi^{-1}(p_j)\), \(j=1,\dots,k\). The following gluing theorem is proved. Let \((d;m_1, \dots, m_k,m)\) be an ample (respectively nef) vector and let \((m; \alpha_1, \dots,\alpha_n) \in\mathbb{Z}_+^{n+1}\) be a nef vector; then \((d;m_1, \dots, m_k,\alpha_1, \dots, \alpha_n)\) is ample (respectively nef). This theorem, combined with the action of the Cremona group on the ample cone gives rise to an algorithmic procedure for detecting new ample classes in the Picard group of rational surfaces. Moreover the author associates a Seshadri-like constant \({\mathcal R}_k({\mathcal L})\), called remainder, to any ample line bundle \({\mathcal L}\) and the integer \(k\). These constants, which are invariant under rescaling of \({\mathcal L}\), allow him to rephrase a classical conjecture of Nagata on the ampleness of \(\pi^* {\mathcal O}_{\mathbb{P}^2} (d)-m\sum^k_{j=1} E_j\) in the following very simple way: \({\mathcal R}_k ({\mathcal O}_{\mathbb{P}^2} (1))=0\) for every \(k\geq 9\). By using the algorithm above the author establishes several results towards Nagata's conjecture and proposes a weaker conjecture on the bound of \({\mathcal R}_k({\mathcal O}_{\mathbb{P}^2}(1))\) in terms of continued fractions approximations of \(\sqrt k\). He also generalizes related results by \textit{Geng Xu} [Manuscr. Math. 86, No. 2, 195-197 (1995; Zbl 0836.14004)] and \textit{O. Küchle} [Math. Ann. 304, 151-155 (1996; Zbl 0834.14024)]. In a final section the meaning of the remainders \({\mathcal R}_k({\mathcal L})\) is explained in terms of volume of a symplectic manifold and the part of it which can be filled by a packing of \(k\) balls.

Recommendations

Cites work

Cited in

(23)

This page was built for publication: Constructing new ample divisors out of old ones

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1974922)