Decompositions into submanifolds that yield generalized manifolds (Q1822098): Difference between revisions

The authors investigate the following interesting problem from decomposition theory: Given an upper semicontinuous decomposition G of an \((n+k)\)-dimensional topological manifold M, let \(X=M/G\) denote the quotient space and \(\pi\) : \(M\to X\) denote the quotient map. Under what conditions is (i) X an ANR ?, (ii) \(\pi\) an approximate fibration ? (Note that an affirmative answer to both (i) and (ii) would imply that X is a generalized manifold, by a result of \textit{R. J. Daverman} and \textit{L. S. Husch} [Mich. Math. J. 31, 197-214 (1984; Zbl 0584.57011)].) The first results along these lines were obtained at the end of the 1970's in a series of papers by \textit{D. S. Coram} and \textit{P. F. Duvall} jun., who also introduced the concept of an approximate fibration [Rocky Mt. J. Math. 7, 275-288 (1977; Zbl 0367.55019)]. The problem was reopened few years later when L. S. Husch, V. T. Liem, and the authors contributed several important results. In particular, the cases when k is 1 or 2 of the general problem are now fairly well understood. The paper under review provides several examples of decompositions of \((n+k)\)-dimensional manifolds, with \(k>2\), such that X fails to be a generalized manifold. On the positive side, they develop a fundamental result that X always has integral cohomological dimension equal to k: Theorem 1. Let X be a complete separable metric space and let \(f: X\to Y\) be a proper map such that each set \(f^{-1}(y)\) has the shape of a closed, connected orientable n-manifold. Then \(\dim_{{\mathbb{Z}}}Y\leq \dim_{{\mathbb{Z}}}X-n\). Next, the authors prove some results which yield answers to questions (i) and (ii) above, e.g. the following Theorem 2. Suppose M is an \((n+k)\)-manifold, \(n\geq k\geq 3\), and suppose G is an upper semicontinuous decomposition of M into compacta having the shape of closed, orientable n-manifolds satisfying: (1) \(\tilde H_ i(g; {\mathbb{Z}})\cong 0\) for all \(g\in G\) and \(i\in \{0,1,...,k-1\}\), and (2) \(\dim M/G<\infty\). Then M/G is a generalized k-manifold. Theorem 3. Suppose G is an upper semicontinuous decomposition of an \((n+k)\)-manifold M, \(n\geq k\geq 2\), into compacta, all having the shape of some fixed n-manifold N such that \(\pi_ 1(N)\) is Hopfian, \({\mathbb{Z}}[\pi_ 1(N)]\) is Noetherian and \(\tilde H_ i(N; {\mathbb{Z}})\cong 0\) for all \(i\in \{0,1,...,k-1\}\). In addition, suppose that the quotient space \(X=M/G\) is finite dimensional. Then the quotient map \(\pi: M\to X\) is an approximate fibration. (In particular, X is a generalized k- manifold.) \{Reviewer's remark: Recently, \textit{R. J. Daverman} has made some further progress in this area - he has determined a large class of manifolds N which are codimension k fibrators, i.e., N is such a closed n-manifold that whenever G is an upper semicontinuous decomposition of an arbitrary topological \((n+k)\)-manifold M such that for every \(g\in G\), \(Sh g=Sh N\) and dim M/G\(<\infty\), then \(\pi: M\to M/G\) is an approximate fibration [''Submanifold decompositions that induce approximate fibrations'', preprint, Knoxville, August 1987].\}