Selections, paraconvexity and PF-normality (Q2448760)

As a generalization of his convex-valued selection theorem, \textit{E. Michael} [Ann. Math. (2) 63, 361--382 (1956; Zbl 0071.15902), Math. Scand. 7, 372--376 (1960; Zbl 0093.12001)] established a selection theorem for paraconvex-valued mappings defined on paracompact spaces. The author proved a similar theorem for collectionwise normal domains in [\textit{N. R. Loufouma Makala}, Topology Appl. 159, No. 1, 153--157 (2012; Zbl 1232.54024)]. In the paper under review, the author proves a PF-normal version of Michael's paraconvex-valued selection theorem. Here, for a number \(\alpha\) with \(0\leq \alpha \leq 1\), a subset \(P\) of a Banach space \(Y\) is said to be \(\alpha\)-paraconvex if, whenever \(p\in Y\) and \(r> 0\) satisfy \(d (p, P) < r\), then \(d (q, P) \leq \alpha r\) for all \(q \in \text{conv} (B(p,r) \cap P)\), where \(d\) is the metric induced by the norm of \(Y\), \(B(p,r)\) denotes the open \(r\)-ball centered at \(x\), and \(\text{conv}(A)\) denotes the convex hull of \(A\). The set \(P\) is said to be paraconvex if it is \(\alpha\)-paraconvex for some \(\alpha < 1\). For an infinite cardinal \(\tau\), a space \(X\) is said to be \(\tau\)-PF-normal if \(X\) is normal and every point-finite open cover \({\mathcal U}\) of \(X\) with \(|{\mathcal U}|\leq \tau\) is refined by a locally finite open cover of \(X\). Let \({\mathcal C}_\alpha(Y)\) denote the set of all compact \(\alpha\)-paraconvex subsets of \(Y\) and \(w(Y)\) the topological weight of \(Y\). The author proves the following theorem: Let \(X\) be a \(\tau\)-PF-normal space, \(Y\) a Banach space with \(w(Y)\leq \tau\), and \(\varphi : X \to {\mathcal C}_\alpha(Y)\) a lower semi-continuous mapping for some \(\alpha <1\). Then \(\varphi\) has a continuous selection. If, for some \(r > 0\), there exists a continuous map \(g : X \to Y\) such that \(d(g(x),\varphi(x)) <r\) for all \(x\in X\), then there exists a continuous selection \(f\) for \(\varphi\) such that \(d(g(x), f (x)) < \delta r\), \(x \in X\), where \(\delta\) is a constant depending only on \(\alpha\). To prove this theorem, the author introduces the concept of Dense Multi-selection Property, which also play the role of a unified approach to paraconvex-valued selection theorems.