Geometry and topology of configuration spaces (Q4520853)

The space of configurations of \(k\) distinct points in a manifold \(M\), \(F_k(M)\), appears and plays an important role in various contexts. The book under review is mainly devoted to a detailed study of the topology of \(F_k(M)\) and its loop spaces (based and free), for \(M=\mathbb{R}^{n+1}\) and \(S^{n+1}\). We are going to describe the main results for Euclidean spaces (the case of spheres being similar, but slightly more complicated). The basic geometric idea is the existence of a natural fibration, \(F_{k+1} (\mathbb{R}^{n+1}) \to F_k(\mathbb{R}^{n+1})\), with fiber a wedge of \(k\) spheres, \(\bigvee_k S^n\). As a consequence, \(F_k (\mathbb{R}^{n+1})\) (respectively the based loop space, \(\Omega F_k(\mathbb{R}^{n+1}))\) is topologically an iterated (nontrivial) twisted product of simple spaces, namely wedges, \(W\), of \(n\)-spheres \((\Omega W\), respectively). The twisting is encoded by the so-called (infinitesimal) Yang-Baxter relations, \((I)YB\), which also appear in quantum group theory, see [\textit{V. G. Drinfel'd}, Leningr. Math. J. 2, No. 4, 829-860 (1991); translation from Algebra Anal. 2, No. 4, 149-181 (1990; Zbl 0718.16034)], and in the theory of finite-type invariants of braids and links, see [\textit{T. Kohno}, Contemp. Math. 179, 123-138 (1994; Zbl 0876.57009)]. The authors use the above approach to derive a complete description of the cohomology algebras, \(H^*F_k(\mathbb{R}^{n+1})\). The defining algebra relations are obtained, by duality, from the \(IYB\) relations, and the result is independent of \(n\) (modulo mild rescaling). Explicit (minimal) cell decompositions of \(F_k(\mathbb{R}^{n+1})\) are also obtained. At the level of homotopy groups, the discussion naturally breaks into two cases: \(n=1\), and \(n>1\). The spaces \(F_k(\mathbb{R}^2)\) are aspheric, and \(YB\) relations are used to exhibit a group presentation for \(\pi_1F_k (\mathbb{R}^2)\). In the 1-connected case \((n>1)\), the authors obtain a complete description of the Hopf algebra \(H_*\Omega F_k (\mathbb{R}^{n+1})\) (and consequently, of the rational homotopy Lie algebra, \(\pi_* \Omega F_k(\mathbb{R}^{n+1}) \otimes\mathbb{Q})\). Modulo rescaling, they find out that \(H_*\Omega F_k(\mathbb{R}^{n+1})\) is isomorphic to the universal enveloping algebra of \(L_k\), \(UL_k\), where \(L_k\) is `the universal \(IYB\) Lie algebra on \(k\) strings'. That is, \(L_k\) is the associated graded Lie algebra (with respect to the lower central series) of \(\pi_1F_k(\mathbb{R}^2)\), the pure braid group on \(k\) strings; see [\textit{T. Kohno}, Invent. Math. 82, 57-75 (1985; Zbl 0574.55009)], for the computation of \(L_k\) with \(\mathbb{Q}\)-coefficients (subsequently extended over \(\mathbb{Z}\) by a number of authors). This intriguing rescaling formula, \(H_*\Omega F_k(\mathbb{R}^{n+1})=U\text{gr} \pi_1F_k(\mathbb{R}^2)\), has been found independently by \textit{F. R. Cohen} and \textit{S. Gitler} [preprint, 1998]. An interesting direction of generalization has been recently proposed by \textit{D. C. Cohen}, \textit{F. R. Cohen} and \textit{M. Xicotencatl} [preprint, math.AT/0005091]. They replace \(F_k(\mathbb{R}^2)\) by the complement of an arrangement, \({\mathcal A}\), of complex hyperplanes, which is assumed to be fiber-type (that is, to possess an iterated fibered structure, like the one of \(F_k (\mathbb{R}^2))\). They go on replacing \(F_k(\mathbb{R}^{n+1})\) above by the complement of an associated arrangement, \({\mathcal A}^q\), of complex subspaces of codimension \(q\), and prove that the rescaling formula still holds. As expected, the computation of free loop space homology, \(H_*\Lambda F_k(\mathbb{R}^{n+1})\), \(n>1\), turns out to be a more difficult problem. In the book under review, the authors attack it by using a variety of additional tools (models of James reduced product type and Adams-Hilton type, Serre and Eilenberg-Moore spectral sequences). Their main results here are: the computation of the mod 2 Poincaré series of \(\Lambda F_3(\mathbb{R}^{n+1})\); the equality cup-length \((H^*(\Lambda F_k (\mathbb{R}^{n+1}); \mathbb{Z}_2))= \infty\), and the related result that \(\Lambda F_k (\mathbb{R}^{n+1})\) contains compact subsets of arbitrarily high Lyusternik-Shnirel'man category. The result on LS-category is finally used to study certain Hamiltonian systems of \(k\)-body type.