Nef divisors for moduli spaces of complexes with compact support (Q522674)

This paper provides a generalization of the result of \textit{A. Bayer} and \textit{E. Macrì} [J. Am. Math. Soc. 27, No. 3, 707--752 (2014; Zbl 1314.14020)] relating the Bridgeland stability manifold and the movable cone of the moduli space of stable objects on smooth projective varieties to the setting of arbitrary separated schemes of finite type. One of the key ideas is, as stated in the section 1.2, that one should introduce a proper definition of the stability function. For a separated scheme \(Y\) of finite type, the authors define a variant \(K^{\text{num}}_c(Y)\) of the numerical Grothendieck group as the quotient of \(K(D_c(Y))\) by the radical of the Euler pairing with perfect complexes on \(Y\), where \(D_c(Y)\) is the full subcategory of objects with proper support in the bounded derived category. Then they introduce the notion of numerical Bridgeland stability condition for compact support on \(Y\) as a pair \(\sigma=(Z_{\sigma}, \mathcal{P}_{\sigma})\) of a group homomorphism \(Z_{\sigma}:K^{\text{num}}_c(Y) \to \mathbb{C}\) and a slicing \(\mathcal{P}_{\sigma}\) of \(D_c(Y)\). The main statement is Theorem 1.2.1, where a family of nef divisors on moduli spaces of stable objects in the above sense is constructed. In the proof, the authors construct a family of t-structures with left-compact support, which is introduced in Definition 2.1.3 as a replacement of the ordinary compact support property so that it behaves well under the derived restriction. This paper also constructs stability conditions in the case when \(Y\) is a smooth scheme, projective over an affine scheme, equipped with a tilting bundle \(E\). In this setting, denoting by \(A\) the endomorphism algebra of \(E^{\vee}\), we have stability conditions \(\sigma_\theta\) on the derived category of finite \(A\)-modules parametrized by stability parameter \(\theta\) for \(A\)-modules in the sense of King. By the tilting equivalence one can construct stability conditions on \(D_c(Y)\) from \(\sigma_\theta\). Then Theorem 1.4.1 says that the numerical divisor class constructed by Theorem 1.2.1 is equivalent to the polarizing ample line bundle on the moduli space of stable \(A\)-modules. The main result of this paper seems to have a wide range of applications as indicated in the section 1.5. The reviewer recommends this paper especially for those interested in some application of Bridgeland stability conditions to birational geometry of moduli spaces and geometric representation theory.