Inequalities of Chern classes on nonsingular projective \(n\)-folds with ample canonical or anti-canonical line bundles (Q6158159): Difference between revisions

The ``geographical'' problem of describing the distribution of Chern numbers of algebraic manifolds of general type was first considered by \textit{I. Reider} [Ann. Math. (2) 127, No. 2, 309--316 (1988; Zbl 0663.14010)] for complex surfaces. After that many authors studied possible generalizations both to the positive characteristic case and to higher dimensions. Studying the geography of complex projective threefolds, B. Hunt posed the problem about the boundedness of the region described by the Chern ratios [\textit{B. Hunt}, J. Differ. Geom. 30, No. 1, 51--153 (1989; Zbl 0712.14022)]. In fact the problem of establishing whether the quotients of the Chern numbers are in a bounded set or not is the basic one and to this end, inequalities involving the Chern classes are crucial. For instance, for surfaces, the classical Noether's inequality and the famous Miyaoka-Yau inequality combined together show that \(c_2/c_1^2\) is bounded. Let \(X\) be a nonsingular projective \(n\)-fold (\(n \geq 2\)) which is either of general type with ample canonical bundle \(K_X\), or Fano, over an algebraically closed field \(\kappa\) of any characteristic. In the paper under review, the authors use a new method to obtain several inequalities in terms of all the Chern classes \(c_1, c_2, \dots , c_n\) of \(X\). This is done by pulling back Schubert classes in the Chow group of the appropriate Grassmannian under the Gauss map of the embedding of \(X\) in \(\mathbb P^N\) defined by a suitable multiple of \(K_X\) or \(-K_X\). The main result is the following. Suppose that \(K_X\) (or \(-K_X\)) is ample. If the characteristic of \(\kappa\) is \(0\) or the characteristic of \(\kappa\) is positive and \(\mathcal O_X(K_X)\) (\(\mathcal O_X(-K_X)\), respectively) is globally generated, then the Chern ratios \((\frac{c_2c_1^{n-2}}{c_1^n},\frac{c_2^2c_1^{n-4}}{c_1^n}, \dots , \frac{c_n}{c_1^n})\) are contained in a convex polyhedron depending on the dimension of \(X\) only. In particular, this gives an affirmative answer to the Hunt question, generalized to all dimensions and to any algebraically closed field. As a corollary, the authors obtain that there exist constants \(d_1, d_2, d_3\) and \(d_4\) depending only on \(n\) such that \(d_1K_X^n \leq \chi_{\mathrm{top}}(X) \leq d_2K_X^n\) and \(d_3K_X^n \leq \chi(X, \mathcal O_X) \leq d_4K_X^n\).