On effective non-vanishing of Weil divisors on algebraic surfaces (Q1774263): Difference between revisions

The ``effective non-vanishing'' conjecture says that for a complete normal variety \(X\), an effective \(\mathbb{R}\)-divisor \(B\) on \(X\) such that \((X,B)\) is Kawamata log terminal, and a nef divisor \(D\) such that \(D-(K_X+B)\) nef and big, \(H^0(X,D)\neq 0\) holds. \textit{Y. Kawamata} [Asian J. Math. 4, No.1, 173--181 (2000; Zbl 1060.14505)] proved this conjecture in the case of log surfaces with only Kawamata log terminal singularities. In the paper under review, the author investigates the related problem when \(D\) is a \(\mathbb{Q}\)-Cartier divisor. Unfortunately, the conjecture doesn't hold under this conditions even in the case of surfaces. The author gives a counterexample by blowing up \(\mathbb{P}^2\) and contracting to cyclic quotient singularities. He obtains a log del Pezzo surface \(X\), where \(D=-K_X\) such that \(6D\) has a nontrivial section but no smaller multiple of \(D\) does. In the course of the computation, he proves a plurigenus formula for log terminal surfaces. However, Kawamata's result can be recovered by strengthening the other conditions. The author proves that for a Gorenstein log del Pezzo surface \(X\) of Picard rank 1 and \(D\) a nef and big \({\mathbb Q}\)-Cartier divisor, then \(H^0(X,D)\neq 0\).