Oscillation revisited (Q1679593)

Let \((X,d)\) and \((Y,\rho)\) be metric spaces and let \(\mathcal{P}_0(X)\) be the family of all non-empty subsets of \(X\). In the paper under review the authors work with the following notion of oscillation of a function \(f:X\to Y\) at a set \(A\in\mathcal{P}_0(X)\), which generalizes the classical notion of the oscillation of a function at a point. For each \(n\in\mathbb{N}\) define \[ \Omega_n(f,A)=\sup\{ \rho(f(x),f(w)): x,w\in S_d(A,\frac{1}{n}) \text{ and } d(x,w)<\frac{1}{n}\}, \] where \[ S_d(A,\varepsilon)=\bigcup_{a\in A}S_d(a,\varepsilon), \] and \(S(a,\varepsilon)\) denotes the open \(d\)-ball of radius \(\varepsilon\) about \(a\in X\). Then \[ \Omega(f,A)=\lim_{n\to\infty}\Omega_n(f,A) \] is called the \textit{oscillation of \(f\) at \(A\)}. The main results proved in the article: (1) Suppose \(C(X,Y)\) is equipped with the topology of uniform convergence and \(\mathcal{P}_0(X)\) is equipped with the locally finite topology. Then the function \(\Omega:C(X,Y)\times\mathcal{P}_0(X)\to [0,+\infty]\) is continuous. In particular, for every continuous function \(f:X\to Y\) the function \(\Omega_f:\mathcal{P}_0(X)\to [0,+\infty]\) defined by \(\Omega_f(A)=\Omega(f,A)\) is continuous. (2) A set \(A\in\mathcal{P}_0(X)\) is called a \textit{\(UC\)-subset} if each continuous function on \(X\) has oscillation zero at \(A\). Then for any \(A\in\mathcal{P}_0(X)\) the following conditions are equivalent: \newline (i) \(A\) is a \(UC\)-subset; \newline (ii) for each \(f\in C(X,\mathbb{R})\), \(\Omega(f,A)\) is finite; \newline (iii) if \(C,E\in\mathcal{P}_0(X)\) are closed, \(C\subset A\) and \(C\cap E=\emptyset\), then the gap between \(C\) and \(E\) is positive.