On homologically locally connected spaces (Q2415946)

The authors consider paracompact Hausdorff spaces \(X\). If \(G\) is a group, then \(H_n(X;G)\) denotes the reduced \(n\)th singular homology group of \(X\). The writers work with the following concept that was introduced by the second author: A closed subset \(A\subset X\) is said to be homologically \(UV^n(G)\) if every neighborhood \(U\) of \(A\) in \(X\) contains another neighborhood \(V\) of \(X\) such that the inclusion of \(V\) into \(U\) induces trivial homomorphisms \(H_k(V;G)\to H_k(U;G)\), \(0\leq k\leq n\). It is noted that if a closed subset \(A\) of an \(\mathrm{ANR}\) is homologically \(UV^n(G)\), then any embedding of \(A\) as a closed subset of an \(\mathrm{ANR}\) is also homologically \(UV^n(G)\). A surjective map \(f:X\to Y\) is called a homologically \(UV^n\)-map if all its fibers \(f^{-1}(y)\) are homologically \(UV^n(G)\). They then define a space \(X\) to be homologically locally connected in dimension \(n\) with respect to \(G\) if the identity map of \(X\) is a \(UV^n(G)\)-map. In this setting, \(X\) is called an \(lc_G^n\)-space. The notion of an algebraic \(\mathrm{ANR}\) is covered in Section 2 -- we will not provide this definition. In this section, the authors obtain some properties of algebraic \(\mathrm{ANR}\)s concerning the ``existence and extensions of homotopies between close chain morphisms.'' Section 3 deals with closed homologically \(UV^n\)-surjections. The ``well known properties of closed \(UV^n\)-maps (concerning extensions of partial realizations and approximate lifting of maps) have chain morphisms analogues for homologically \(UV^n\)-maps. We also provide some Dugundji type extensions and approximate extensions of chain morphisms. Obviously, all results established in Section 3 are valid for \(lc_G^n\)-spaces.'' Section 4 involves a characterization of \(lc_G^n\)-spaces in terms of the existence of homotopies between any close chain morphisms (Proposition 4.1). Two questions are discussed as well. Question 1.1. Is any (do they mean every) metric \(n\)-dimensional \(lc_G^n\)-space an absolute neighborhood algebraic retract? Question 1.2. Let \(f:Y\to X\) be a closed \(UV^n(G)\)-surjection between metrizable spaces. Is it true that \(X\) is an \(lc_G^n\)-space?