Factoring continuous mappings defined on subspaces of topological products (Q2217236)

Let \(Y\) be a subset of a product \(X = \Pi_{i\in I} X_i\) of a family \(\{X_i : i \in I\}\) of sets, \(Z\) be a set, and \(f : Y \rightarrow Z\) be an arbitrary mapping. For \(a\in X\), the \(\sigma\)-product \(\sigma X(a)\) is the subset of \(X\) containing all points which differ from \(a\) in no more than finitely many coordinates. If \(J\subseteq I\), then \(f\) depends on \(J\) if \(f(x) = f(y)\) for every \(x, y \in Y\) with \(p_J(x) = p_J(y)\), where \(p_J : X \rightarrow \Pi_{i\in J} X_i\) is the projection. When \(f\) depends on \(J\), then there exists a mapping \(g: p_J(Y) \rightarrow Z\) satisfying \(f = g \circ p_J \upharpoonright Y\). For the mapping \(f\), let \(\mathcal{J}(f)\) be the family of all subsets \(J\) of \(I\) such that \(f\) depends on \(J\). The systematic study of the family \(\mathcal{J}(f)\), in the case when \(X\) and \(Z\) are topological spaces and \(f\) is a continuous mapping, was initiated by \textit{W. W. Comfort} and the author of this review in [Sci. Math. Jpn. 69, No. 3, 363--377 (2009; Zbl 1193.54011)]. One of the main questions considered in this paper is the existence of the smallest by inclusion element of \(\mathcal{J}(f)\). We recall that when \(Z\) is Hausdorff and \(\sigma X(a) \subseteq Y\), for some \(a\in X\), the existence of the smallest element in \(\mathcal{J}(f)\) is a well-known fact [loc. cit., Proposition 1.8 and Remark 1.9]. The author of this paper shows in Theorem 2.3 that \(J_f=\bigcap\mathcal{J}(f)\) is the smallest (by inclusion) element of \(\mathcal{J}(f)\) under the assumptions that \(X\) has the Tychonoff product topology, \(Z\) is a \(T_0\)-space, and \(Y\) is a finitely mixing subspace of \(X\). Furthermore, the mapping \(g: p_{J_f}(Y) \rightarrow Z\) satisfying \(f = g \circ p_{J_f} \upharpoonright Y\) is continuous. (We recall [loc. cit., Definition 1.1(f)] that if \(\kappa\) is a cardinal, a subset \(Y\) of \(X\) is called \(\kappa\)-invariant provided for every \(x,y\in Y\) and every \(J\subset I\), with \(|J|<\kappa\), the point \(z \in X\) defined by \(p_J(x)=p_J(z)\) and \(p_{I\setminus J}(y) = p_{I\setminus J}(z)\) is in \(Y\). When \(\kappa = \omega\), the author calls such sets \(Y\) finitely mixing subsets of \(X\).) The author also shows that the same conclusions as in Theorem 2.3 are valid when \(Y\) is an open dense subspace of \(X\) and \(Z\) is Hausdorff (Theorems 2.8) or when \(Y\) is a subspace of \(X\) containing a \(\sigma\)-space \(\sigma X(a)\) for some \(a\in X\) and \(Z\) is a regular \(T_1\)-space (Theorem 2.16). But, if in Theorem 2.16 the space \(Z\) is Hausdorff, then it is possible that \(J_f=\bigcap\mathcal{J}(f)\) is the smallest (by inclusion) element of \(\mathcal{J}(f)\) and the mapping \(g: p_{J_f}(Y) \rightarrow Z\) satisfying \(f = g \circ p_{J_f} \upharpoonright Y\) is not continuous (see Theorem 2.11 and Remark 2.13). Several interesting examples are also given in this paper refuting some possible generalizations or conjectures and two questions are posed.