Bounded cohomology of amenable covers via classifying spaces (Q2032792)

Let \(X\) be a path-connected CW-complex. Denote by \(H^{*}(X;\mathbb{R})\) and \(H_{b}^{*}(X;\mathbb{R})\) the singular and bounded cohomology of \(X\), with \(\mathbb{R}\)-coefficients, respectively. Similarly, denote by \( H_{*}(X;\mathbb{R})\) and \(H_{*}^{\ell^{1}}(X;\mathbb{R})\) the singular and \(\ell^{1}\)-homology of \(X\), with \(\mathbb{R}\)-coefficients. The authors compare these (co-)homology groups using a suitable cover of the space \(X\) and the family of amenable subgroups of the fundamental group of \(X\). A novelty in this work is the application of properties of families of subgroups of a group, in this case they use the family of amenable subgroups of the fundamental group of \(X\). More concretely, an amenable open cover, \(U\), of \(X\) is an open cover by path-connected components such that, for every \(V\in U\) and \(x\in V\), the image of the homomorphism \(i_{*}:\pi_{1}(V,x)\to \pi_{1}(X,x)\) induced by inclusion, is an amenable subgroup of \(\pi_{1}(X,x)\). Moreover, the cover \(U\) is convex if all finite intersections of members of \(U\) are path-connected or empty. The main results are: \textbf{Theorem 1.1. } Let \(X\) be a path connected CW complex and \(U\) be an amenable open cover of \(X\). Let \(N\) be the nerve of \(U\) and \(|N|\) be its geometric realisation. Let \(c_{X}:H_{b}^{*}(X;\mathbb{R})\to H^{*}(X;\mathbb{R})\) be the comparison map. Then, (1) if \(U\) is convex, then the comparison map factors through the nerve map \(\nu:X\to |N|\). That is, \(\nu^{*}\varphi=c_{X}\) for some \(\mathbb{R}\)-linear map \(\varphi: H^{*}_{b}(X;\mathbb{R})\to H^{*}(|N|;\mathbb{R})\), (2) if the multiplicity of \(U\) is at most \(m\), then the comparison map \(c_{X}\) vanishes for all \(*\geq m\). The second result concerns \(\ell^{1}\) homology: \textbf{Theorem 1.2. } Under the hypotheses of Theorem 1.2, let \(c^{\ell^{1}}_{X}: H_{*}(X;\mathbb{R})\to H^{\ell^{1}}_{*}(X;\mathbb{R})\) be the comparison map. Then, (1) if \(U\) is convex, then the \(c^{\ell^{1}}_{X}\) comparison map factors through the nerve map \(\nu:X\to |N|\), that is, there is an \(\mathbb{R}\)-linear map \(\varphi: H_{*}(|N|;\mathbb{R})\to H^{\ell^{1}}_{*}(X;\mathbb{R}))\) such that \(\varphi \nu_{*}= c^{\ell^{1}}_{X}\), (2) if the multiplicity of \(U\) is at most \(m\), then the comparison map \( c^{\ell^{1}}_{X}\) vanishes for all \(*\geq m\).