A canonical linear system associated to adjoint divisors in characteristic p>0

From MaRDI portal
Publication:475829



Abstract: Suppose that X is a projective variety over an algebraically closed field of characteristic p>0. Further suppose that L is an ample (or more generally in some sense positive) divisor. We study a natural linear system in |KX+L|. We further generalize this to incorporate a boundary divisor Delta. We show that these subsystems behave like the global sections associated to multiplier ideals, H0(X,mJ(X,Delta)ensorL) in characteristic zero. In particular, we show that these systems are in many cases base-point-free. While the original proof utilized Kawamata-Viehweg vanishing and variants of multiplier ideals, our proof uses test ideals.


Let \(X\) be a normal variety over an \(F\)-finite field of positive characteristic and \(\Delta\) an \(\mathbb Q\)-divisor such that \(K_X + \Delta\) is \(\mathbb Q\)-Cartier. In characteristic zero one can associate the multiplier ideal \(\mathcal J(X, \Delta)\) while in positive characteristic one can associate the test ideal \(\tau(X, \Delta)\). Let \(L\) be an ample divisor. The paper under review study certain linear system in \(|K_X + L|\) which behaves like the global section associated to multiplier ideal \(H^0(X, \mathcal J(X, \Delta) \otimes L)\). He proved analogous results in positive characteristic using test ideal. In particular, in corollary 4.5, he proved that if \(L - K_X - \Delta\) is big and nef then \(\tau(X,\Delta) \otimes_{\mathcal O_X} (L + nM)\) is globally generated for all globally generated ample Cartier divisor \(M\). As mentioned by the author, the strategy of the proof is similar to that of [\textit{D. S. Keeler}, Am. J. Math. 130, No. 5, 1327--1336 (2008; Zbl 1159.14003)]. He also improved the bound in the case of curves.




Cited in
(27)








This page was built for publication: A canonical linear system associated to adjoint divisors in characteristic \(p>0\)

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q475829)