Equivariant Moore spaces. II: The low-dimensional case (Q2266480): Difference between revisions

This paper is a continuation of the author's [Algebraic and geometric topology, Lect. Notes Math. 1126, 238-270 (1985)]. In this paper, he considers spaces X having the following characteristics: (1) \(\pi_ 1X=\pi\); (2) \(H_ 0(X;{\mathbb{Z}}\pi)={\mathbb{Z}}\); (3) \(H_ i(X;{\mathbb{Z}}\pi)=0\), \(i\neq 2\), \(i\leq k\); (4) \(H_ 2(X;{\mathbb{Z}}\pi)=M\), as right \({\mathbb{Z}}\pi\)-modules. Let us say such a space has type (M,2,k;\(\pi)\). If X has type (M,2,k;\(\pi)\) for all k, then we say it has type (M,2;\(\pi)\) and call it (or its universal cover) an equivariant Moore space. The author proves the following: If X has type (\({\mathbb{Z}},2,4;\pi)\) and \(\pi\) is such that \(H^ 3(\pi;{\mathbb{Z}})\neq 0\) and has no 2-torsion, then the first k-invariant of X (in \(H^ 3(X;{\mathbb{Z}}))\) is 0. This gives a topological condition which may be violated by purely algebraically given data. That is, with \(\pi\) as above, there may exist a projective \({\mathbb{Z}}\pi\)-chain complex C satisfying the homological conditions above but having non-zero first homological k-invariant. It follows that such a C cannot be chain equivalent to the singular complex of a space - i.e., C is not realizable. A result of independent interest in the paper constructs an equivariant homotopy equivalence between Eilenberg and MacLane's bar and W-constructions. It follows that the Eilenberg-MacLane model for Eilenberg-MacLane spaces is equivariantly homotopy equivalent to that of \textit{R. J. Milgram} [Ill. J. Math. 11, 242-250 (1967; Zbl 0152.405)].