On the stable classification of spin four-manifolds (Q1423466)

This paper gives the stable and weak stable classification of smooth and smooth spin \(4\)-manifolds with finitely presented fundamental groups satisfying \(H_2(\pi;{\mathbb Z}_2)=0\) and \(H_2(\pi;{\mathbb Z}_3)=0\). Recall that two smooth \(4\)-manifolds are said to be stably equivalent if they become diffeomorphic after connected summing with enough copies of \(S^2\times S^2\) or \(S^2\widetilde\times S^2\). Two spin \(4\)-manifolds are said to be weak stably equivalent if they become spin diffeomorphic after connected summing with enough copies of \(S^2\times S^2\) and the K3 surface. The result in the simply connected case goes back to work of Wall. In that case, stable equivalence is classified by the signature, because once two \(4\)-manifolds have the same signature, they become cobordant. Given a cobordism one can cancel extra zero- and five-handles, and trade one- and four-handles for two- and three-handles to arrive at a cobordism with only two- and three-handles. Each two-handle is attached along a homotopically trivial loop, so corresponds to taking the connected sum with \(S^2\times S^2\) or \(S^2\widetilde\times S^2\). Turning the cobordism upside down turns the three-handles into two-handles, so the middle level of the cobordism is obtained from either end by taking a suitable connected sum. In the presence of a spin structure or non-trivial fundamental group one must pay attention to the attaching maps for the two handles. This is what the author of the paper under review does. Any \(4\)-manifold with fundamental group \(\pi\) comes with a natural map into the classifying space \(B\pi\) corresponding to the universal cover. This pair (manifold, map) represents an element of the cobordism group \(\Omega_4(B\pi)\). There is a natural map from the cobordism group into \(H_4(B\pi)\) taking a manifold and map to the push-forward of the fundamental map. To analyze the kernel of this map in the spin case, the author applies the Atiyah-Hirzebruch spectral sequence. This is the point where the homological assumptions on the fundamental group are needed. The paper also includes some nice corollaries and examples of the main theorem.