Varieties with pseudoeffective canonical divisor and their Kodaira dimension
Gilberto Bini
TL;DR
This paper studies Mori's conjecture that a smooth complex projective variety $X$ is uniruled precisely when its Kodaira dimension $kod(X)$ is negative. It develops a framework connecting $kod(X)$ to Castelnuovo-Mumford regularity of multiples of $K_X$ via Koszul cohomology, using a generic complete intersection curve $C\subset X$ and a spectral sequence argument to translate cohomological data into regularity bounds. Under the assumption that $K_X$ is pseudoeffective, the authors prove that $kod(X)\ge0$ by establishing CM-regularity bounds $reg_A((1-s)K_X)\ge d+1$ (for $s\ge s_A$) and extracting a nonvanishing Koszul obstruction, thereby ruling out negative Kodaira dimension in dimension $d\ge3$. Together with the BDPP nonvanishing result, this yields that $kod(X)=-\infty$ iff $X$ is uniruled in dimension $d\ge3$, with reductions to curves and surfaces for smaller dimensions. The work clarifies the nonvanishing direction of Mori's conjecture under a precise KO-spectral framework and highlights the role of Green-Lazarsfeld-type Koszul cohomology in birational geometry.
Abstract
Let X be a smooth, projective variety over the field of complex numbers. Here we focus on a conjecture attributed to Shigefumi Mori, which claims that X is uniruled if and only if the Kodaira dimension of X is negative.
