A generalization of fiber-type arrangements and a new deformation method (Q1292652)

This article introduces the important new class of hypersolvable arrangements, a common generalization of fiber-type (supersolvable) and 3-generic arrangements which shares many of their most pleasant topological properties. An arrangement \(\mathcal A\) is hypersolvable if there is a sequence \({\mathcal A}_0 \subset {\mathcal A}_1 \subset \cdots\subset {\mathcal A}_n={\mathcal A}\) of subarrangements in which \({\mathcal A}_i\) is solvable in \({\mathcal A}_{i+1}\) for each \(i\). An inclusion \({\mathcal B}\subset {\mathcal A}\) is solvable if \(\mathcal B\) is line-closed in \(\mathcal A\) and the intersection of every pair of hyperplanes of \({\mathcal A}-{\mathcal B}\) is contained in a hyperplane of \({\mathcal B}\). (\(\mathcal B\) is line-closed if \(K\in {\mathcal B}\) for any \(K\in {\mathcal A}\) containing \(H\cap H'\) for \(H, H'\in {\mathcal B}\).) A further condition ensures that a solvable extension increases rank by at most one. The rank increases precisely when \({\mathcal B}\) is a flat in \(\mathcal A\), in which case \(\mathcal B\) is modular in \(\mathcal A\). Thus \(\mathcal A\) is supersolvable if and only if each solvable extension in some (equivalently, any) such sequence increases the rank, i.e. iff \(n=\text{rank}({\mathcal A})\). In case \({\mathcal B}\subseteq {\mathcal A}\) is a solvable extension with \(\text{rank}({\mathcal A})=\text{rank}({\mathcal B})\) the authors give a geometric construction of a lift \(\widetilde{\mathcal A}\) of \(\mathcal A\), \(\text{rank}(\widetilde{\mathcal A})= \text{rank}({\mathcal A}) +1\), in which \(\mathcal B\) is modular, and which has the same flats of ranks one and two as \(\mathcal A\). The proof of the latter fact requires Desargues' Theorem in projective 3-space. Using this construction, the authors extend many of the most well-known properties of fiber-type arrangements to this new class, with the Orlik-Solomon algebra \(A({\mathcal A})\) replaced by its quadratic closure \(\overline{A}({\mathcal A})\). The latter algebra depends only on flats of rank less than three, and agrees with \(A({\mathcal A})\) when \(\mathcal A\) is fiber-type. The fundamental group of the complement \(M({\mathcal A})\) is determined by a generic 3-dimensional section, and thus is the same for \(\mathcal A\) and the lift \(\widetilde{\mathcal A}\) . So, if \(\mathcal A\) is hypersolvable, the Poincaré polynomial of \(\overline{A}({\mathcal A})\) splits over the integers, with roots equal to \(|{\mathcal A}_i - {\mathcal A}_{i-1}|\), and the fundamental group of \(M({\mathcal A})\) is an iterated semidirect product of free groups via automorphisms which are homologically trivial (called ``almost direct products''). There is a ``quadratic LCS formula'' relating the ranks of lower central series quotients of \(\pi_1(M({\mathcal A}))\) to the Poincaré polynomial of the quadratic algebra \(\overline{A}({\mathcal A})\). Also, it is shown that the quadratic algebra \(\overline{A}({\mathcal A})\) is Koszul, and that the associated graded of the integral holonomy Lie algebra of \(\mathcal A\) is isomorphic to the graded \({\mathbb Z}\)-Lie algebra associated with the lower central series of \(\pi_1(M({\mathcal A}))\). In recent work the authors have extended their deformation technique to show that any hypersolvable arrangement \(\mathcal A\) is a section of a supersolvable arrangement \(\widetilde{\mathcal A}\), \(\text{rank}(\widetilde{\mathcal A})=n\), with the same flats of ranks one and two [Topol. Appl. 118, 103-111 (2002)]. This yields simplified proofs of most of the results in the paper under review, since the complement of \(\widetilde{\mathcal A}\) is a \(K(\pi,1)\) for \(\pi=\pi_1(M({\mathcal A}))\). In addition, the stronger deformation result shows that \(\mathcal A\) is a formal arrangement if and only if \({\mathcal A}\) is supersolvable. Thus the only free or \(K(\pi,1)\) arrangements in the hypersolvable class are the supersolvable ones. This observation provides negative answers to Questions 8.3 and 8.4 of the paper under review.