Quillen's plus construction and the D(2) problem (Q2389455)

Wall's D(2) problem is the following: Let \(X\) be a finite 3-complex with \(H^3(X;\beta)=0\) for all coefficient bundles \(\beta\). Is \(X\) homotopy equivalent to a finite 2-complex? In the present paper, given a finite 3-complex \(X\) with cohomological dimension 2, it is shown that a complex homotopy equivalent to \(X\) can be constructed by applying Quillen's plus construction to the Cayley complex of a finite group presentation. This reduces the D(2) problem to a question about perfect normal subgroups. More precisely, let \(\varepsilon=\langle g_1,\dots,g_n\mid R_1,\dots,R_m\rangle\) be a finite presentation for a group \(E\). We say that a normal subgroup of \(E\) is finitely closed when it is the normal closure in \(E\) of a finitely generated subgroup. Let \(K\vartriangleleft E\) be finitely closed and perfect. Let \({\mathcal K}_\varepsilon\) denote the Cayley complex associated to \(\varepsilon\). Then by a Theorem of Quillen [see \textit{J. Rosenberg}, Algebraic \(K\)-theory and its applications. Graduate Texts in Math. 147. New York, NY: Springer-Verlag (1994; Zbl 0801.19001)], there is a 3-complex \({\mathcal K}_\varepsilon^+\), containing \({\mathcal K}_\varepsilon\) as a subcomplex which is characterized up to homotopy equivalence by the following two properties: (i) the inclusion \({\mathcal K}_\varepsilon\to{\mathcal K}_\varepsilon^+\) induces the quotient map \(E\to E/K\) on fundamental groups; (ii) \(H_*({\mathcal K}_\varepsilon^+,{\mathcal K}_\varepsilon;M)=0\) for all left modules over \(\mathbb Z[E/K]\). In the present paper \({\mathcal K}_\varepsilon^+\) is constructed explicitly, using the fact that \(K\) is finitely closed in order to ensure that we end up with a finite cell complex. The main result of the paper (Theorem 3.4) is the following: Let \(X\) be a finite connected 3-complex with \(H^3(X;\beta)=0\) for all coefficient bundles \(\beta\). Then \(X\) has the homotopy type of \({\mathcal K}_\varepsilon^+\) for some finite presentation \(\varepsilon\) of a group \(E\), where the plus construction is taken with respect to some perfect finitely closed normal subgroup \(K\vartriangleleft E\). From this Theorem the author proves (Theorem 4.2) that Wall's D(2) problem is equivalent to the following: Let \(K\) be a perfect finitely closed normal subgroup of a finitely presented group \(E\). For each finite presentation \(\varepsilon\) of \(E\), there exists a finite presentation \({\mathcal G}\) of \(E/K\), such that we have a chain homotopy equivalence over \(\mathbb Z[E/K]: C_*({\mathcal K}_\varepsilon)\bigotimes_E\mathbb Z[E/K]\to C_*({\mathcal K}_{{\mathcal G}})\).