A characterization of high order freeness for product arrangements and answers to Holm's questions (Q2041921)

In the paper under review, the authors study the so-called \(m\)-freeness of hyperplane arrangements. Let \(\mathbb{K}\) be a field of characteristic zero and let \(V\) be an \(l\)-dimensional vector space over \(\mathbb{K}\). A central hyperplane arrangement \(\mathcal{A}\) is a finite collection of hyperplanes in \(V\) such that each contains the origin. We call \(\mathcal{A}\) an \(l\)-arrangement when we emphasize the dimension of \(V\). Denote by \(Q \in S:= \mathbb{K}[x_{1},\dots, x_{l}]\) a defining polynomial of \(\mathcal{A}\). For \(\mathbf{a} = (a_{1},\dots, a_{l})\in \mathbb{N}^{l}\) one uses multi-index notations, namely \[ |\mathbf{a}|= a_{1} + \dots + a_{l}, \quad \mathbf{a}! = a_{1}! \cdots a_{l}!, \quad \text{and} \quad \partial^{\mathbf{a}} = \partial_{1}^{a_{1}} \cdots \partial_{l}^{a_{l}}, \] where \(\partial_{i} := \frac{\partial}{\partial \, x_{i}}\). Define \[ D^{(m)}(S) = \bigoplus_{|\mathbf{a}| = m}S \cdot \partial^{\mathbf{a}} \] for \(m\geq 1\) and \(D^{(0)}(S) = S\). One can observe that \(D^{(m)}(S)\) is an \(S\)-submodule of \(\mathrm{End}_{\mathbb{K}}(S)\). For a central hyperplane arrangement \(\mathcal{A}\) we define an \(S\)-submodule \(D^{(m)}(\mathcal{A})\) of \(D^{(m)}(S)\) by \[ D^{(m)}(\mathcal{A}) = \{ \theta \in D^{(m)}(S) \, | \, \theta(QS) \subset QS\}. \] We call \(D^{(m)}(\mathcal{A})\) the module of \(m\)-th order \(\mathcal{A}\)-differential operators. Finally, we say that \(\mathcal{A}\) is \(m\)-free if \(D^{(m)}(\mathcal{A})\) is a free \(S\)-module. Since \(QD^{(m)}(S) \subseteq D^{(m)}(\mathcal{A}) \subset D^{(m)}(S)\), the rank of \(D^{(m)}(\mathcal{A})\) is \(s = s_{m}(l) = \binom{l+m-1}{m}\) provided that \(\mathcal{A}\) is \(m\)-free. We say that \(\theta = \sum_{|\mathbf{a}|=m}f_{\mathbf{a}}\partial^{\mathbf{a}} \in D^{(m)}(S)\) is homogeneous of degree \(j\) and write \(\deg (\theta) = j\) if \(f_{\mathbf{a}}\) is zero or homogeneous of degree \(j\) for each \(\mathbf{a}\). If \(m\geq 1\) and if \(\mathcal{A}\) is \(m\)-free with a homogeneous basis \(\{\theta_{1},\dots, \theta_{s}\}\), we define \(m\)-exponents by the multi-set \(\exp_{m}(\mathcal{A}) = \{ \deg (\theta_{1}), \dots, \deg (\theta_{s})\}\). In the context of further developments on the \(m\)-freeness, one can ask the following questions (proposed by P. Holm around 2002): Q1) Does \(m\)-free imply \((m+1)\)-free for any arrangement? Q2) Are all arrangements \(m\)-free for \(m\) large enough? It turns out, however, that the answers to the above two questions are negative, and this is the main core of the present paper under review. Theorem A. Let \((\mathcal{A}_{1}, V_{1})\) and \((\mathcal{A}_{2}, V_{2})\) be arrangements with \(\dim V_{1} > 0\) and \(\dim V_{2} > 0\). The following conditions are equivalent: a) \((\mathcal{A}_{1} \times \mathcal{A}_{2}, V_{1} \oplus V_{2})\) is \(m\)-free. b) Both \((\mathcal{A}_{1}, V_{1})\) and \((\mathcal{A}_{2}, V_{2})\) are \(i\)-free for all \(1 \leq i \leq m\). c) \((\mathcal{A}_{1} \times \mathcal{A}_{2}, V_{1} \oplus V_{2})\) is \(i\)-free for all \(1\leq i \leq m\). This result implies that the answer to Q2 is no. Example. Let \(\mathcal{A}\) and \(\mathcal{A}'\) be two arrangements. If \(\mathcal{A}\) is not \(1\)-free, then the product arrangement \(\mathcal{A} \times \mathcal{A}'\) is not \(m\)-free for any \(m\geq 1\). In particular, a generic arrangement is known to be not \(1\)-free and hence if \(\mathcal{A}\) is generic and \(\mathcal{A}'\) is arbitrary, then \(\mathcal{A} \times \mathcal{A}'\) is not \(m\)-free for any \(m\geq 1\). For \(l\geq 2\) we define \(\mathrm{Shi}_{l}\) as an \((l+1)\)-arrangement defined by \[ Q(\mathrm{Shi}_{l}) = z \prod_{i=1}^{l}x_{i}(x_{i}-z) \prod_{1\leq i < j\leq l}(x_{i}-x_{j})(x_{i}-x_{j}-z). \] Theorem B. The arrangement \(\mathrm{Shi}_{l}\) is not \(2\)-free for all \(l\geq 2\). Since \(\mathrm{Shi}_{l}\) is known to be \(1\)-free thus the answer to Q1 is also no.