The structure of normal surfaces (Q1078271): Difference between revisions

Let Y be a normal Moishezon surface i.e. it has two independent meromorphic functions. (This is equivalent to the assumption that any resolution of Y is projective.) Let \(K_ Y\) be a canonical divisor of Y and let \(R(Y)=\oplus_{m\geq 0}H^ 0(Y,{\mathcal O}_ Y(mK_ Y))\) be the canonical ring. The Kodaira dimension \(\kappa(Y)\) is defined as (transcendence degree of \(R(Y))-1.\) If \(R(Y)={\mathbb{C}}\), then \(\kappa(Y)=- \infty\) by convention. The author had proved earlier [Duke Math. J. 51, 877-888 (1984)] that minimal models exist in the category of normal surfaces. This paper classifies minimal normal surfaces according to their Kodaira dimensions and describes their structure when \(\kappa\leq 1\). The tables in the theorems are too long to be reproduced here. Just as an example, the author proves that even if \(K_ Y\) is numerically effective, \(\kappa(Y)\) may still be \(-\infty\), unlike in the non-singular case. The proofs depend (amongst many other things) on the construction and study of a suitable ''irredundant'' non-singular model S of Y with a divisor D on S.