Diagonal form of the Varchenko matrices (Q1631383)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Diagonal form of the Varchenko matrices |
scientific article |
Statements
Diagonal form of the Varchenko matrices (English)
0 references
6 December 2018
0 references
Let \(\mathcal{A}= \{H_{1}, \dots, H_{N}\} \subset \mathbb{R}^{d}\) be a hyperplane arrangement. For \(B \subset I = \{1, \dots,N\}\) we define \(H_{B} = \bigcap_{i \in B}H_{i}\). We say that \(\mathcal{A}\) is a general arrangement in \(\mathbb{R}^{d}\) if for any subset \(B\subseteq I\), the cardinality \(|B| \leq d\) implies \(\dim H_{B} = d - |B|\) while \(|B| > d\) implies that \(H_{B} = \emptyset\). If for all \(B \subseteq I\) with \(H_{B} \neq \emptyset\), we have \(\dim H_{B} = d - |B|\), then \(\mathcal{A}\) is called semigeneral in \(\mathbb{R}^{d}\). Denote by \(L(\mathcal{A})\) the intersection poset of \(\mathcal{A}\). We denote the set of regions by \(\mathcal{R}(\mathcal{A})\) and \(r(\mathcal{A}) := |\mathcal{R}(\mathcal{A})|\). Now for each hyperplane \(H_{i}\) we assign an indeterminate \(x_{i}\) and for each pair or regions \((R_{i},R_{j})\) of \(\mathcal{A}\), we set \[ \text{sep}(R_{i},R_{j}) = \{ H_{a} \in \mathcal{A} : H_{a} \text{ separates } R_{i} \text{ and } R_{j}\}. \] The Varchenko matrix for \(\mathcal{A}\), denoted by \(V(\mathcal{A}) = [V_{ij}]\), is the \(r(\mathcal{A}) \times r(\mathcal{A})\) matrix with rows and columns indexed by \(\mathcal{R}(\mathcal{A})\) and entries \[ V_{ij} = \prod_{H_{a} \in \text{sep}(R_{i},R_{j})} x_{a}. \] If for \((R_{i},R_{j})\) there is no separating hyperplane, then \(V_{i,j}=1\). To each element \(M \in L(\mathcal{A})\), then we assign \(x_{M} = \prod_{M \subseteq H_{a}}x_{a}\). For a hyperplane arrangement \(\mathcal{A}\) we denote by \(\chi_{A}\) its characteristic polynomial, and for \(M \in L(\mathcal{A})\) we define \[ \mathcal{A}_{M} = \{H \in \mathcal{A} : M \subseteq H\}, \] \[ \mathcal{A}^{M} = \{ M \cap H \neq \emptyset: H \in \mathcal{A} \setminus \mathcal{A}_{M}\}. \] Theorem (Varchenko). Let \(\mathcal{A}=\{H_{1}, \dots, H_{N}\}\) be a real affine finite hyperplane arrangement. Let \(M \in L(\mathcal{A})\) and we set \(n(M) = r(\mathcal{A}^{M})\), \(p(M) = \bigg|\frac{d}{dt} \chi_{\mathcal{A}_{M}}(t)_{t=1}\bigg|\). Then the determinant of the Varchenko matrix is given by \[ \det \, V(\mathcal{A}) = \prod_{M \in L(\mathcal{A}), M \neq \mathbb{R}^{d}}(1-x^{2}_{M})^{n(M)p(M)}. \] Main result. Let \(\mathcal{A} = \{H_{1}, \dots, H_{N}\}\) be a real affine finite hyperplane arrangement in \(\mathbb{R}^{d}\). Assign an indeterminate \(x_{a}\) to each \(H_{a}\), \(a \in I = \{1, \dots, N\}\). Then the Varchenko matrix \(V(\mathcal{A})\) has a diagonal form over \(\mathbb{Z}[x_{1}, \dots, x_{N}]\) if and only if \(\mathcal{A}\) is in semigeneral position. In that case, the diagonal entries of the diagonal form of \(V(\mathcal{A})\) are exactly the products \(\prod_{a\in B}(1-x_{a}^{2})\) ranging over all \(B \subset I\) such that \(H_{B} \in L(\mathcal{A})\).
0 references
hyperplane arrangement
0 references
Varchenko matrix
0 references
diagonal form
0 references