A generic slice of the moduli space of line arrangements (Q723160)

In the paper under review, the authors study the compactification of the locus parametrizing lines having a fixed intersection with a given line, insider the moduli space of line arrangements with weights in the complex projective plane. Let us recall basic notions. Given an arrangement of \((n+1)\) lines \(\{\ell_{1},\dots,\ell_{n+1}\}\) in the complex projective plane, we label the line \(\ell_{n+1}\) as \(\ell_{A}\). Configurations of \((n+1)\) label lines in \(\mathbb{P}^{2}\) up to projective equivalence are parametrized by the open moduli space \(M(\mathbb{P}^{2}, n+1)\) which has a family of geometric compactifications \(\overline{M}_{\overline{\beta}}(\mathbb{P}^{2},n+1)\) depending on a weight vector \(\overline{\beta} = (\beta_{1},\dots, \beta_{n+1})\). The weight domain of possible weights \(\overline{\beta}\) is \[ \mathcal{D}(3,n+1) = \bigg\{ \overline{\beta} \in \mathbb{Q}^{n+1} : \sum_{i=1}^{n+1} \beta_{i} > 3, \, 0 < \beta_{i} \leq 1 \bigg\}. \] In general, these compactifications are not irreducible. From now on we assume that \(\ell_{A}\) will always have multiplicity \(1\) and thus we will denote our weight set \(\beta \in \mathcal{D}(3,n+1)\) as \((\overline{\omega},1)\). We will consider a restriction morphism \[ \phi_{A} : M_{(\overline{\omega},1)} (\mathbb{P}^{2},n+1) \rightarrow M_{0, \overline{\omega}} \] induced naturally by considering the intersection of \(\ell_{A}\) with the lines \(\ell_{i}\) where \(i \in \{1,\dots,n\}\). Next, we take the fiber of this restriction over a generic point \(q \in M_{0,\overline{\omega}}\) and then take closure of this fiber in the compact moduli space of weighted hyperplane arrangements. Definiton 1. Let \(q \in M_{0,\overline{\omega}} \subset \overline{M}_{0,\overline{\omega}}\) be a generic point. We define \(R_{\overline{\omega}}(q)\) as the closure in \(\overline{M}_{\overline{\omega},1}(\mathbb{P}^{2},n+1)\) of the fiber product of the following diagram: \[ \begin{tikzcd} R_{\vec\omega}(q) \arrow{r}\arrow{d} & \overline M_{(\vec\omega,1)}(\mathbb P^2,n+1) \\ q \arrow{r} & \overline M_{0,\vec\omega} \arrow{u}\end{tikzcd}. \] It turns out that for certain weights \(\overline{\omega}_{0}\) one has \(R_{\overline{\omega}_{0}}(q) \cong \mathbb{P}^{n-3}\) -- we will call such weights as admissible weights. Definiton 2. Let \(\overline{\omega}_{0} = (\omega_{0_{1}},\dots,\omega_{0_{n}})\) be a set of rational numbers such that for every subset \(I \subseteq \{1, \dots, n\}\) the inequality \(\sum_{i \in I} \omega_{0_{i}} \leq 2\) holds. The set of admissible weights is \[ \mathcal{D}_{n}^{R} = \bigg\{ (\omega_{1},\dots, \omega_{n}) \in \mathbb{Q}^{n} : 1 \geq \omega_{i} > 0 : \sum_{i=1}^{n} \omega_{i} \geq 2, \omega_{i} \geq \omega_{0_{i}}\bigg\}. \] Now we can formulate the main results of the paper (all results are obviously considered over the complex numbers). Theorem 1. For weights \(\overline{\omega }\) in the set of admissible weights \(\mathcal{D}^{R}_{n}\) and generic choice of \(q \in M_{0, \overline{\omega}}\), the locus \(R_{\overline{w}}(q)\) is smooth with normal crossing boundary and there are birational morphisms \[ R_{\overline{\omega}}(q) \rightarrow \overline{M}_{0,\overline{\omega}} \rightarrow \mathbb{P}^{n-3} \] . Theorem 2. For a generic choice of \(q\) and \(n\geq 5\), there do not exist weights \(\overline{\omega}\) such that \(R_{\overline{\omega}}(q) \cong \overline{M}_{0,n}\). The next main result is devoted to Chow quotients which are used to study the geometry of moduli spaces. Given the collection of \(n\) points \(p_{i}\) in the dual projective space \(\widehat{\mathbb{P}}^{2}\) such that the point \(p_{i}\) is dual to the line \(\ell_{i}\), we consider the locus, in an appropriate Chow variety, that parametrizes the cycles associated to the orbits \(\overline{G\cdot(p_{1},\dots,p_{n})}\) where \(G \subset \text{SL}(3, \mathbb{C})\) is the group that fixes the intersection of the associated lines \(\ell_{i}\) with \(\ell_{A}\). By normalizing the closure of this locus in the Chow variety, one recovers \(R_{\overline{\omega}}(q)\) with \(\overline{\omega} =(1,\dots,1) :=1^{n}\). Theorem 3. For a generic choice of \(q\), the space \(R_{1^{n}}(q)\) is isomorphic to the normalization of a closed subvariety of the Chow quotient \((\hat{\mathbb{P}}^{2})^{n} /\!/_{\text{Ch}} G\) where \(G \subset \text{SL}(3,\mathbb{C})\) is the group fixing the line \(\ell_{A}\) pointwise. The paper is extremely nicely written and pleasant to read.