On nef subvarieties (Q2319880)

Fixing \(q\geq 0\), the line bundle \(\mathscr L\) on the projective variety \(X\) is \(q\)-ample if for all coherent \(\mathscr F\) there exists \(m_0\) such that \(H^i(X;{\mathscr F}\otimes{\mathscr L}^{\otimes m})=0\) for all \(m\geq m_0\) and \(iq\). For example \(0\)-ampleness is the same as the usual ampleness. [\textit{B. Totaro}, J. Eur. Math. Soc. (JEMS) 15, No. 3, 731--754 (2013; Zbl 1277.14007)] is the standard reference. [\textit{J. C. Ottem}, Adv. Math. 229, No. 5, 2868--2887 (2012; Zbl 1285.14006) ] defined ampleness of subvarieties that are not necessarily complete intersection via this partial ampleness property on the blow-up. The paper under review brings contributions to both the general theory of \(q\)-ampleness for divisors, and to the study of positivity for subvarieties. The author starts with two versions of Fujita vanishing. A particular case is that if \(\mathscr L\) is \(q\)-ample, and \(\mathscr F\) is coherent, then there exists \(M0\) such that \(H^i(X;{\mathscr F}\otimes {\mathscr L}^{\otimes m}\otimes \mathscr P)=0\) for all \(iq\), \(m\geq M\), and all nef line bundles \(\mathscr P\). The proofs rely on techniques introduced by Arapura and Totaro, studying resolutions of the diagonal. The second part of the paper is devoted to applications to positive subvarieties. The author defines nefness for subvarieties by asking that the exceptional \(E\) be close to being \(r-1\)-ample. He studies some of the geometric properties of nef subvarieties, and proves transitivity for ampleness and nefness: If \(Z\subset Y\) is ample/nef and \(Y\subset X\) is ample/nef, then \(Z\subset X\) is ample/nef. This is a nontrivial application of his generalizations of Fujita vanishing. Another application is an extension of a result of [\textit{J.-P. Demailly} et al., Isr. Math. Conf. Proc. 9, 165--198 (1996; Zbl 0859.14005) ]: If \(Y\subset X\) is an ample subscheme of codimension \(r\) and the line bundle \(\mathscr L\) on \(X\) has the property that \({\mathscr L}|_Y\) is \(q\)-ample, then \(\mathscr L\) itself is \(q+r\)-ample. In the future, this could be applied in inductive arguments. An important invariant in birational geometry is the movable cone of (Weil) divisors. It can be defined as the cone generated by classes of divisors \(\pi_*D\), where \(\pi:Y\to X\) is dominant, and \(D\) is nef on \(Y\). Similarly, the author defines a weakly movable cone \(\mathrm{Wmov}_d(X)\) in the space of numerical cycle classes \(N_d(X)\). He finds an inclusion \(\mathrm{Mov}_d(X)\subseteq\mathrm{Wmov}_d(X)\), where the former is the movable cone of cycle classes of [\textit{M. Fulger} and \textit{B. Lehmann}, Algebr. Geom. 4, No. 1, 1--28 (2017; Zbl 1376.14009)]. He further proves that the two cones have similar properties, and asks whether they are equal. The author employs a diverse array of nontrivial cohomological and geometric techniques in his proofs that are not widely used, so there may be more applications.