Hypertetrahedral arrangements (Q2126079)

Let \(k\) be an algebraically closed field of characteristic zero, \(S = k[x_{0}, \dots, x_{n}]\). For a homogeneous polynomial \(f \in S\) of degree \(d\), we denote by \(J_{f}\) the Jacobian ideal generated by the partial derivatives \(\partial_{x_{j}}\,f\) with respect to \(x_{j}\) with \(j \in \{0, \dots, n\}\). For any graded \(S\)-module \(M\), we denote by \(\mathrm{indeg}(M)\) the initial degree of \(M\), that is, the minimum degree of a non-zero element in \(M\). Moreover, we denote by \(\mathrm{Der}_{k}(S)\) the free \(S\)-module of rank \(n+1\) generated by the partial derivatives \(\partial_{x_{j}}\). A hyperplane arrangement \(\mathcal{A} = \{H_{1}, \dots, H_{m}\} \subset \mathbb{P}^{n}_{k}\) is a collection of \(m\) distinct hyperplanes of \(\mathbb{P}^{n}_{k}\). The module \(\mathrm{Der}(-\log \, \mathcal{A})\) of logarithmic derivations of \(\mathcal{A}\) is the set of \(S\)-linear derivations \(\theta \in \mathrm{Der}_{k}(S)\) such that \(\theta(Q) \subset (Q)\), where \(Q\) is the defining equation of \(\mathcal{A}\). We say that \(\mathcal{A} \subset \mathbb{P}^{n}_{k}\) is free if \(\mathrm{Der}(-\log \, \mathcal{A})\) is a free \(S\)-module of rank \(n+1\). In the paper under review the authors study a new class of arrangements, introduced by the authors therein, called hypertetrahedral arrangements. Let us present briefly their construction. Denote by \(e_{0}=(1: 0 : 0 : \, \dots \, : 0)\), \dots \(e_{n}=(0 : 0 : \, \dots \, : 1)\) the vertices of an \(n\)-dimensional simplex. Let \(\mathcal{L}_{ij} \subset \mathbb{P}^{n}_{k}\) the codimension \(2\) linear subspace passing through the vertices \(e_{0}, \dots, \widehat{e_{i}}, \dots, \widehat{e_{j}}, \dots, e_{n}\) with \(i<j\). A hypertetrahedral arrangement is a hyperplane arrangement \(\mathcal{A}\) such that any of its hyperplanes passes through at least one linear subspace \(\mathcal{L}_{i,j}\) with \(0 \leq i < j \leq n\). The hypertetrahedral arrangement \(\mathcal{A}\) is complete if it contains the hyperplanes \(L_{i} : \, x_{i}=0\) for \(i \in \{0, \dots, n\}\). One denotes by \(\mathcal{H}(s_{i,j})_{0 \leq i < j \leq n}\) the set of all complete hypertetrahedral arrangements \(\mathcal{A}\) in \(\mathbb{P}^{n}_{k}\) with \(s_{i,j}+2\) different hyperplanes passing through \(\mathcal{L}_{i,j}\). Let \(L(\mathcal{A})\) be the intersection lattice and \(X \in L(\mathcal{A})\) be a linear subspace. We define \(W_{X}\) as the intersection, of minimal dimension, of coordinate hyperplanes containing \(X\). If \(X\) is not contained in any coordinate hyperplane, then one sets \(W_{X} = \emptyset\) and then \(X\) is called inner. We denote additionally by \(\mathcal{A}_{W_{X}}\) the localized arrangement with respect to \(W_{X} \in L(\mathcal{A})\). Let \(X \in L(\mathcal{A})\) be a linear subspace such that \(X \subsetneq W_{X}\) and we assume additionally that \(\mathrm{supp}(\mathcal{A}_{X} \setminus \mathcal{A}_{W_{X}}) = \{x_{i_{0}}, \dots, x_{i_{k}}\}\). We define \(\Gamma_{X} = (V_{X}, E_{X})\) the graph associated to \(X\) with vertices \(V_{X} = \{i_{0}, \dots, i_{k}\}\) and \(E_{X}\) the set of edges \((i_{l},i_{m})\) such that there is \(H \in \mathcal{A}_{X} - \mathcal{A}_{W_{X}}\) with \(\mathrm{supp}(H) = \{ x_{i_{l}}, x_{i_{m}}\}\) (here it means that \(H : \, x_{i_{l}} - \lambda x_{i_{m}} = 0\) for \(\lambda \neq 0\)). It is known that complete hypertetrahedral arrangements are, in general, not free, and one of the main goals of the paper is to characterize whether these arrangements are locally free. Theorem 1. A hypertetrahedral arrangement \(\mathcal{A}\) is locally free if and only if for each \(X \in L(\mathcal{A})\), the complete hypertetrahedral arrangement \(\mathcal{A}_{W_{X}}\) is free and \(\Gamma_{X}\) is chordal. Moreover, if \(n\geq 3\) and \(\mathcal{A} \in \mathcal{H}(s_{i,j})\) is general, then \(\mathcal{A}\) is not locally free. We say that a hypertetrahedral arrangement is general if any intersection outside the \(n\)-th simplex has minimal dimension. The next goal of the paper is to establish lower and upper bounds for the initial degree of the syzygy module associated with any complete hypertetrahedral arrangement in terms of the values \(s_{i,j}\). First result into this direction is devoted to hypertetrahedral arrangement with \(n=2\) -- these are called shortly as triangular arrangements. Proposition 1. Let \(\mathcal{A} \in \mathcal{H}(s_{i,j})\) be a complete general triangular arrangement with \(s_{0,1} \leq s_{0,2} \leq s_{0,3}\). Then \(\mathrm{indeg}(\mathcal{J}_{\mathcal{A}}) = s_{0,1} + s_{0,2} + 1\). Moreover, if \(s_{0,3} \geq s_{0,1} + s_{0,2}\), then one has \[\mathrm{Der}(-\log \, \mathcal{A})_{d} \subset S_{d-1} \theta_{E}\] for all \(1 \leq d \leq \min (s_{0,1}+s_{0,2}, s_{0,3})\), where \(\theta_{E}\) denotes the Euler derivation. The main general result in the indicated direction, quite technical in its nature, needs a preparation. Fix integers \(i_{0},q_{0}\) with \(0 \leq i_{0} < q_{0} \leq n\). One denotes by \(T^{i_{0},q_{0}}\) the set of all possible \((n-1)\)-tuples \(((i_{0},j_{0},q_{0}), \dots, (i_{n-2},j_{n-2},q_{n-2}))\) of triples \((i_{m},j_{m},q_{m})\) of integers with \(0 \leq i_{m},j_{m},q_{m} \leq n\) such that \(i _{0} \not\in \{i_{0},q_{0}\}\) and for each \(1 \leq m \leq n-2\), the following two conditions are satisfied: i) \(j_{m} \not\in \{i_{0}, \dots, i_{m-1},j_{0}, \dots, j_{m-1}, q_{0}, \dots, q_{m-1}\}\), ii) \(i_{m}< q_{m}\) and \(i_{m},q_{m} \in \{i_{0}, \dots, i_{m-1}, j_{0}, \dots, j_{m-1}, q_{0}, \dots, q_{m-1}\}\). Each \((n-1)\)-tuple \(v \in T^{i_{0},q_{0}}\) uniquely determines a set \[S_{v}^{i_{0},q_{0}} := \{s_{i_{0},q_{0}}, s_{i_{0},j_{0}}+s_{j_{0},q_{0}},s_{i_{1},j_{1}}+s_{j_{1},q_{1}}, \dots, s_{i_{n-2},j_{n-2}}+s_{j_{n-2},q_{n-2}}\},\] and one defines \(m_{v}^{i_{0},q_{0}} := \min \, S_{v}^{i_{0},q_{0}}\) and \(M^{i_{0},q_{0}} = \max_{v \in T^{i_{0},q_{0}}}\{m_{v}^{i_{0},q_{0}}\}\). Theorem 2. Fix an integer \(n\geq 2\) and let \(\mathcal{A} \in \mathcal{H}(s_{i,j})\). Let \(D:= \max_{0 \leq i_{0} < q_{0} \leq n} \{M^{i_{0},q_{0}}\}\). Then for all \(1\leq d \leq D\) one has \[\mathrm{Der}(-\log \, \mathcal{A})_{d} \subset S_{d-1}\theta_{E}.\]