Linear systems on toric varieties (Q1362551)

A version of Fujita's conjectures is proven for \(\mathbb{Q}\)-Gorenstein toric varieties. More precisely: Theorem. Let \(X\) be a projective \(\mathbb{Q}\)-Gorenstein toric variety of dimension \(n\), and let \(L\in \text{Pic}(X)\) be ample. (i) Let \(t>n\). Then every positive multiple of \(K_X+tL\) that is Cartier, is base point free. (ii) Let \(t>n+1\). Then every positive multiple of \(K_X+tL\) that is Cartier, is very ample. Moreover, if \((X,L)\neq (\mathbb{P}^n,{\mathcal O}(1))\), then (i) and (ii) hold with \(\geq\)-signs. The proof of this theorem uses toric Mori theory, as developed by \textit{R. Reid} [in: Arithmetic and geometry, Pap. dedic. Shafarevich, Vol. II, Prog. Math. 36, 395-418 (1983; Zbl 0571.14020)]. Note that since toric Mori theory is properly combinatorial, the theorem holds over any base field. A key ingredient in the proof is an estimate on the length of extremal rays of the cone of curves: If \(X\) is as above, then the length of any extremal ray is \(\leq n+1\), with strict inequality if \(X\neq\mathbb{P}^n\) (in Mori theory, this is proven for arbitrary smooth varieties).