Divisionally free arrangements of hyperplanes (Q284634)

The author delivers several significant (and really exciting) results around the notion of freeness for hyperplane arrangements. Let \(V\) be an \(\ell\)-dimensional vector space over an arbitrary field \(\mathbb K\), \(S=\mathrm{Sym}(V^\ast)=\mathbb K[x_1,\dots,x_\ell]\) and \(\mathrm{Der }S:=\oplus_{i=1}^\ell S\partial_{x_i}\). A hyperplane arrangement \(\mathcal A\) is a finite set of hyperplanes in \(V\). We say that \(\mathcal A\) is central if every hyperplane is linear and in this case we fix a linear form \(\alpha_H\in V^\ast\) such that \(\mathrm{ker}(\alpha_H)=H\). We assume here that all hyperplane arrangements are central. An \(\ell\)-arrangement is an arrangement in an \(\ell\)-dimensional vector space. For \(\mathcal A\) we denote by \(L(\mathcal A):=\{\cap_{H\in\mathcal B}H:\mathcal B\subset\mathcal A\}\) the intersection lattice. It is well known that \(L(\mathcal A)\) has a partial order by reverse inclusion. Let \(L_i(\mathcal A):=\{X\in L(\mathcal A)\mid \mathrm{codim }X=i\}\). For \(X\in L(\mathcal A)\) we define the localization \(\mathcal A_X\) of \(A\) at \(X\) by \(\mathcal A_X:=\{H\in\mathcal A:X \subset H\}\), which is a subarrangement of \(\mathcal A\). Moreover, we define the restriction \(\mathcal A^X\) of \(\mathcal A\) onto \(X\) by \(\mathcal A^X=\{H\cap X:H\in\mathcal A\setminus\mathcal A_X\}\), which is an arrangement in \(X\simeq\mathbb K^{\dim X}\). For a hyperplane arrangement \(\mathcal A\) we define the logarithmic derivation module \[ \mathcal D(\mathcal A)=\{\theta\in\mathrm{Der}S \mid\theta(\alpha_H)\in S\cdot\alpha_H,\forall\,H\in\mathcal A\}. \] Now we say that \(\mathcal A\) is free with exponents \(\exp(\mathcal A)=(d_1,\dots,d_\ell)\) if \(\mathcal D(\mathcal A)\) is generated as an \(S\)-module by \(S\)-independent generators \(\theta_1,\dots,\theta_\ell\) with \(\deg\theta_i=d_i\) for every \(i\). The study of free arrangements was initiated by \textit{H. Terao} [Proc. Japan Acad., Ser. A 56, 389--392 (1980; Zbl 0476.14016); J. Fac. Sci., Univ. Tokyo, Sect. I A 27, 293--312 (1980; Zbl 0509.14006)] around 36 years ago and it is still really difficult to determine the freeness. Let us here recall some important results around the freeness from [loc. cit.], i.e., the addition-deletion theorem and the factorization theorem. The addition-deletion theorem allows us to construct free arrangements. For a central arrangement \(\mathcal A\) and \(H\in\mathcal A\) we define \(\mathcal A':=\mathcal A\setminus H\). Then we call \((\mathcal A,\mathcal A',\mathcal A^H)\) a triple. Theorem (Addition-deletion). Let \((\mathcal A,\mathcal A',\mathcal A^H)\) be a triple. Then any two of the following three imply the third: (a) \(\mathcal A\) is free with \(\exp(\mathcal A)=(d_1,\dots,d_{\ell-1},d_\ell)\). b) \(\mathcal A'\) is free with \(\exp(\mathcal A)=(d_1,\dots,d_{\ell-1},d_\ell-1)\). c) \(\mathcal A^H\) is free with \(\exp(\mathcal A^H)=(d_1,\dots,d_{\ell-1})\). Now for a given poset \(L(\mathcal A)\) we define the Möbius function \(\mu: L(\mathcal A)\rightarrow\mathbb Z\) by \(\mu(X)=1\) if \(X=V\) and \(\mu(X)=-\sum_{X \subsetneq Y\in L(\mathcal A)}\mu(Y)\) if \(X \neq V\). Then the characteristic polynomial \(\chi(\mathcal A, t)\) of \(\mathcal A\) is defined as \[ \chi(\mathcal A,t)=\sum_{X \in L(\mathcal A)}\mu(X)t^{\dim X}. \] Theorem (Factorization). If \(\mathcal A\) is free with \(\exp(\mathcal A)=(d_1,\dots,d_\ell)\), then \(\chi(\mathcal A,t)=\prod_{i=1}^\ell(t-d_i)\). The famous deletion-restriction formulae tells us that for a triple \((\mathcal A,\mathcal A',\mathcal A^H)\) one has \[ \chi(\mathcal A,t)=\chi(\mathcal A',t) - \chi(\mathcal A^H,t). \] Now let us observe that both the factorization theorem and the deletion-restriction formulae decodes some information about the divisibility of characteristics polynomials for triplets. In the article the author starts deeper investigations around divisibility properties of characteristic polynomials in the context of freeness. The main result of the paper is the following beautiful division theorem. Theorem 1. Let \(\mathcal A\) be a central \(\ell\)-arrangement. Assume that there is a hyperplane \(H\in\mathcal A\) such that \(\chi(\mathcal A^H,t)\) divides \(\chi(\mathcal A,t)\) and \(\mathcal A^H\) is free. Then \(\mathcal A\) is free. As it is observed, Theorem 1 allows to check the freeness of \(\mathcal A\) by constructing a divisional tower. Theorem 2. An \(\ell\)-arrangement \(\mathcal A\) is free if there is a flag \[ V=X_0\supset X_1\supset\dots\supset X_{\ell-2} \] such that \(X_i\in L_i(\mathcal A)\) for \(i=0,\dots,\ell-2\) and \(\chi(\mathcal A^{X_{i+1}},t)|\chi(\mathcal A^{X_i},t)\) for \(i=0,\dots,\ell-3\). We call this flag a divisional flag. The key advantage of Theorem 2 is that it provides a purely combinatorial method to check the freeness. The celebrated conjecture due to Terao tells us that the freeness is a combinatorial property, and this conjecture is still open for \(3\)-arrangements. It is an extremely interesting (and hard) problem to determine new relations which would allow us to prove this conjecture. One way to approach this conjecture is to find a class of free arrangements such that whether an arrangement belongs to that class or not can be determined by combinatorics. If we could find such a class and show that all free arrangements belong to it, then Terao's conjecture is true. For that purpose Terao introduced the so-called inductively free arrangements \(\mathcal{IF}\) and he showed the properties written above. In the paper the author improves Terao's approach in the following sense. Definition. We say that an \(\ell\)-arrangement \(\mathcal A\) is divisionally free if \(\mathcal A\) has a divisional flag as in Theorem 2. We denote by \(\mathcal{DF}_\ell\) the set of all divisionally free \(\ell\)-arrangements and \(\mathcal{DF}:=\bigcup_{\ell \geq 1}\mathcal{DF}_\ell\). Theorem 3. If \(\mathcal A\in\mathcal{DF}\), then \(\mathcal A\) is free. Moreover, \(\mathcal{IF}\subset\mathcal{DF}\) and this containment is strict, and whether \(\mathcal A\in\mathcal{DF}\) or not depends only on combinatorics, i.e., on \(L(\mathcal A)\). Moreover, the author shows that recursively free arrangements at the hill (see Section 5 for the definition) are divisionally free. The last result, which we would like to report here, is the following statement on the so-called \(k\)-extended Shi arrangements (see Section 6 for definition). Theorem 4. The \(k\)-extended Shi arrangements are divisionally free. Hence the freeness of \(k\)-extended Shi arrangements depend only on their combinatorics. Let us point out here that the paper contains also some new tools and technics which can be further developed, for instance the non-trivial Localization and Remainder Theorem. In order to show these results the author uses a large spectrum of methods, for instance the theory of multiarrangements (which does not appear a priori in the statements of the results), Ziegler multiplicities and restrictions etc. Moreover, the paper is extremely friendly written (there are some small typos) and it is pleasant to read, even though that there are some really technical parts.