On homological stability for configuration spaces on closed background manifolds (Q273525)

It is well-known that for manifolds \(M\) with boundary, manifolds with a puncture or with an end, the configuration space \(C_k(M)\) of \(k\) unordered points in \(M\) embeds in \(C_{k+1}(M)\) (by ``expanding'' and adding a point near the boundary, the puncture or at ``infinity''), and that this embedding induces a homology isomorphism through a range of degrees that depends on \(k\). This is called homological stability and the range over which this isomorphism holds has been computed by several authors for various ring coefficients. For closed manifolds \(M\) stabilization maps are not known to exist, which leaves open the question of when the homology of configuration spaces on closed manifolds stabilises. This question is investigated in much detail and in various ways in this paper. The introduction of this paper is quite exhaustive. The first part of this paper compares configuration spaces and section spaces. The scanning map first investigated by Dusa McDuff for general smooth manifolds sends \(C_k(M)\) to \(\Gamma_c(\dot{T}M)_k\) the space of compactly supported sections of the fiberwise one-point compactification of the tangent bundle of \(M\) having degree \(k\). The degree of such sections is defined in \S2 and it is shown how to compute it. It turns out that section spaces when localized at a prime \(p\) can be compared and are homotopy equivalent under certain conditions depending on \(\chi (M)\) and \(p\) [\textit{M. Bendesky} and \textit{J. Miller}, Q. J. Math. 65, No. 3, 807--815 (2014; Zbl 1306.55010)]. The authors extend this result and prove (Theorem A) that for the maximal range where \(H_*(C_k(M))\) and \(H_*(\Gamma_c(\dot{T}M)_k)\) are isomorphic (i.e. in the \textit{stable range}), \(H_*(C_k(M);\mathbb Z)\cong H_*(C_{k+2}(M);\mathbb Z)\) if \(M\) is odd dimensional (with a refinement if \(\dim M=3,7\)), and \(H_*(C_k(M);\mathbb Z\left[{1\over 2}\right])\cong H_*(C_{k+2}(M);\mathbb Z\left[{1\over 2}\right])\). The statement is more complicated for even dimensional \(M\) and involves \(\chi (M)\). Yet this comparison theorem is sharp as the authors can check by computing \(H_{n-1}(C_k(S^n);\mathbb Z)\) if \(n\) is even and \(k\) belongs to the stable range with respect to the homological degree \(n-1\) (Proposition 2.12). A second part of this paper defines ``replication maps'' which form a new set of maps between configuration spaces defined whenever \(M\) admits a non-vanishing vector field. These are maps defined (roughly) between deformation retracts \(\rho_r: C_k^\delta (M)\rightarrow C_{rk}^\delta (M)\) by adding \(r-1\) points near each point of the configuration in the direction of the vector field. Theorem B of this paper states that in the presence of such a vector field and if \(r\geq 2\), the replication map \(\rho_r\) induces an isomorphism in homology in the stable range with coefficients in \(\mathbb Z_{(\ell )}\) if \(\ell\) is a set of primes each not dividing \(r\). If \(M\) is not closed, the induced map is always injective. A similar statement for configuration spaces with labels in a fibre bundle \(E\rightarrow M\) is stated (Theorem C). A third part of the paper uses Theorem B in the case of open manifolds \(M\) and a cofibration sequence relating configuration spaces on \(M\) and on \(M\setminus *\) (punctured \(M\)) to show that \(H_*(C_k(M);\mathbb F)\cong H_*(C_{rk}(M);\mathbb F)\) in a range, where \(M\) is closed, smooth and even dimensional, \(\mathbb F\) is a field of positive characteristic \(p\) and \(r\geq 2\) an integer coprime to \(p\) such that \(p\) divides \((\chi -1)(r-1)\). This is Theorem D. Theorem A says that in odd dimensions there are at most two stable integral homologies, depending on the parity of the number of points of \(k\). In even dimensions, even when taking homology with \(\mathbb Z_{(p)}\) coefficients, there may be infinitely many different stable homologies: one for each possible \(p\)-adic valuation of \(2k-\chi\). With fields of odd characteristic \(p\), the authors manage to give estimate on the count of such number of stable homologies (Corollary E). A final useful result proven in this paper (Corollary F) is a periodicity theorem similar to an earlier result of [\textit{R. Nagpal}, FI-modules and the cohomology of modular representations of symmetric groups, {\url arXiv:1505.04294}]. It states that for prime \(p\) and \(M\) smooth even dimensional with \(\chi (M)\neq 0\), \[ H_i(C_k(M);\mathbb F)\cong H_i(C_{k+p^{L(M)}}(M);\mathbb F) \] with field coefficients \(\mathbb F\) of odd characteristic \(p\), \(L(M)\) is one plus the \(p\)-adic valuation of \(\chi (M)\) and \(k>>i\) (ie. the homology is ``eventually'' periodic as \(k\rightarrow\infty\)). This is a paper with several interesting results.