Rational manifold models for duality groups (Q2324614)

The theme of the paper is the following question: given a discrete group \(\Gamma\), construct interesting manifolds (typically not compact) rationally equivalent to the classifying space \(B\Gamma\). Its main theorem in this direction is the following: Theorem 3. Let \(\Gamma\) is a discrete group with finite classifying space \(B\Gamma\) and assume that the twisted coefficient cohomology \(H^*(B\Gamma;\mathbb{Z}\Gamma)\) is concentrated in degree \(d>2\) (i.e.~\(\Gamma\) is a \(d\)-dimensional duality group). Then there is a compact conncected \((d+3)\)-dimensional manifold \(M\) with connected boundary \(\partial M\) and \(\pi_1\)-isomorphisms \(\partial M\to M\to B\Gamma\) such that the rational homoloogy of the universal covering \(\tilde M\) is trivial. Note that all higher homotopy groups of such an \(M\) are torsion groups, the paper therefore calls such manifolds rationally aspherical. Using the reflection group method of Michael Davis, out of the manifolds of Theorem 3 the paper then constructs interesting closed manifolds, and in particular proves the following second main result of the paper: Theorem 4. There is a dimension \(d\ge 7\) and a \(d\)-dimensional \emph{closed} rationally aspherical manifold with non-zero \(L^2\)-Betti number in degrees outside half the dimension. This sheds light on the so-called Singer conjecture which predicts that for a closed manifold with contractible universal covering all \(L^2\)-Betti number in degrees different from half the dimension are zero. The proof proceeds by first constructing from the algebraic information of \(D:=H^d(\Gamma;\mathbb{Z}\Gamma)\) a \(\Gamma\)-CW-complex \(X\) with \(\pi_1(X)=\Gamma\) and equivariant rational homology of \(\tilde X\) equal to \(D\) concentrated in degree \(r\), using tools from rational homotopy theory. Then, it is shown that \(X\to B\Gamma\) is a \((d+r+1)\)-dimensional rational Poincaré duality pair. Next, using surgery and the \(\pi\)-\(\pi\)-theorem, a corresponding manifold pair is constructed, proving Theorem 3. Finally, reflection trick constructions going back to \textit{B. Okun} and \textit{K. Schreve} [Algebr. Geom. Topol. 16, No. 5, 2549--2569 (2016; Zbl 1383.20028)] are used to discuss the application to the Singer conjecture. The construction is a bit inexplicit: if the Singer conjecture holds for manifolds of dimension \(<7\) then the reflection trick construction (with fundamental group \(F_2^4\), the cartesian product of \(4\) copies of the free group on two generators) produces a closed \(7\)-manifold \(M\) with \(b^{(4)}_2(M)\ne 0\). If the Singer conjecture is not true for manifolds of dimension \(\le 7\), then such a counterexample provides also an example for Theorem 4. The very nice paper not only provides a the proofs of the main theorems, it also recalls relevant theory and results from rational homotopy theory; about (Poincaré) duality spaces, pairs, and groups; from Wall's surgery theory; the reflection group method; and about \(L^2\)-Betti numbers and could indeed serve as a (very concise) introduction to all of these subjects. It also suggests a number of intersting questions and directions of further research.