Multi-Gieseker semistability and moduli of quiver sheaves (Q1740707)

The idea of D. Gieseker's construction in geometric invariant theory (GIT) is to construct the moduli space as a quotient of a locally closed subset of a Grassmannian by the action of a general linear group. The semistable sheaves then appear as orbit closures of semistable points with respect to a line bundle on the Grassmannian. Another classical example of the role of GIT in the study of moduli spaces is A. King's construction of moduli spaces of semistable representations of a quiver with a fixed dimension vector. There is a natural notion of semistability, depending on some tuple of real parameters for a quiver representation, and King identifies this notion with D. Mumford's notion of semistability for points on the representation variety by defining a suitable linearization. That is, a parameter needs to be chosen in order to define the appropriate notion of semistability. In the case of sheaves, this parameter is an ample line bundle $L$ on a space $X$ so that the Hilbert polynomials are calculated with respect to this line bundle. In the second case, the parameter is a tuple of real numbers. When Hilbert polynomials are considered as parameters defining semistability, it is natural to fix a tuple $\underline{L}=(L_1,\ldots, L_N)$ of ample line bundles and to compute Hilbert polynomials with respect to $\mathbb{Q}$-line bundles $L_1^{q_1} \otimes \cdots \otimes L_N^{q_N}$ for tuples $(q_1,\ldots, q_N)\in \mathbb{Q}^N_{\geq 0}$. \par There is however a lack of a notion of semistability for irrational parameters. Thus D. Greb, J. Ross, and M. Toma [\textit{D. Greb} et al., Geom. Topol. 20, No. 3, 1539--1610 (2016; Zbl 1400.14032)] introduced the notion of multi-Gieseker semistablity. First, by fixing a tuple $\underline{L}=(L_1,\ldots, L_N)$ of ample line bundles, they consider nonnegative linear combinations $\sum_{j=1}^N \sigma_j P^{L_j}_{E}$, with $\sigma_1,\ldots, \sigma_N \in \mathbb{R}$, of the Hilbert polynomials $P^{L_j}_{E}$ with respect to the $L_j$ of a coherent sheaf $E$. They provide a functorial construction of a moduli space by defining a functor from the category $\text{Coh}(X)$ of coherent sheaves on $X$ to the category of representations of a certain quiver which depends on the line bundles $L_j$ and the section spaces of tensor powers of these line bundles. \par Maslovarić and Seppänen address on whether the construction of multi-Gieseker semistable sheaves has a counterpart for quiver sheaves, i.e., quiver sheaves are triples $(\mathcal{E}_1, \mathcal{E}_2, \varphi)$, where the $\mathcal{E}_i$ are vector bundles on $X$, and $\varphi: \mathcal{E}_1 \to \mathcal{E}_2$ is a morphism of $\mathcal{O}_X$-modules. They construct coarse moduli spaces for semistable quiver sheaves using a functorial method that realizes these as subschemes of moduli spaces of representations of a twisted quiver with relations, and they also show the projectivity of the moduli space in the case when the quiver has no oriented cycles (Theorem 1, page 260): given a quiver $Q$, $k=\overline{k}$ of characteristic $0$, a smooth projective variety $X$ over $k$, and a fixed topological type $\tau$, if the family of $\sigma$-semistable quiver sheaves of topological type $\tau$ is bounded, then the coarse moduli space of $\sigma$-semistable quiver sheaves of topological type $\tau$ exists as a quasi-projective scheme $M^\sigma$ over $k$, and if $Q$ has no oriented cycles and $\sigma$ is rational, $M^\sigma$ is projective over $k$. \par The authors also investigate the parameter dependence of the moduli space (Theorem 3, page 261): for torsion-free quiver sheaves, the set $\sigma$ of symmetric stability conditions is partitioned into equivalence classes by a finite set of rational linear hyperplanes in such a way that the set of $\sigma$-semistable quiver sheaves is constant on these classes, and if $Q$ has no oriented cycles, the moduli spaces $M^\sigma$ are projective over $k$ for arbitrary bounded nonnegative symmetric $\sigma$.