Notes on toric varieties from Mori theoretic viewpoint (Q1876285): Difference between revisions

The main result here is a version of the cone theorem for an \(n\)-dimensional projective toric variety \(X\) with integral toric divisors \(D_1,\dots,D_r\). If \(D=\sum d_j D_j\) is a \({\mathbb Q}\)-divisor such that \(0\leq d_j\leq 1\) for all~\(j\) and \(K_X+D\) is \({\mathbb Q}\)-Cartier, then for each extremal ray \({\mathbb R}_{\geq 0}[C]\) of \(\text{NE}(X)\) there exists an \(n-1\) dimensional cone \(\tau\) of the fan of \(X\) such that \([V(\tau)]\in {\mathbb R}_{\geq 0}[C]\) and \[ -(K_X+D).V(\tau)\leq n+1, \] except when \(X={\mathbb P}^n\) and \(\sum d_j<1\). As a corollary one obtains a strong version of Fujita's conjecture in the toric case, incorporating results of \textit{M. Mustaţa} [Tôhoku Math. J., II. Ser. 54, No. 3, 451--470 (2002; Zbl 1092.14064)] and \textit{R. Laterveer} [Tôhoku Math. J., II. Ser. 48, No. 3, 451--458 (1996; Zbl 0874.14047)]. With \(X\) and \(D\) as above, and \(L\) a line bundle on \(X\), if \(L.C\geq n\) (respectively, \(\geq n+1\)) for every integral curve \(C\subset X\), then \(K_X+D+L\) is nef (respectively ample) unless \(X={\mathbb P}^n\) and \(\sum d_j<1\) and \(L={\mathcal O}(n)\) (respectively \(D=0\) and \(L={\mathcal O}(n+1)\)). The paper also contains a discussion of \({\mathbb Q}\)-factorial toric Fano varieties of Picard number~\(1\), and a proof that every toric variety has a small projective \({\mathbb Q}\)-factorialisation.