No label defined (Q4949136)

It seems, that the author wants to avoid the notion of continuous convergence for nets of functions, but implicitly he uses this notion. For topological spaces \(X,Y\) denote by \(Y^X\) the set of all functions from \(X\) to \(Y\), and by \(C(X,Y)\) the set of all continuous functions respectively. (0) If \((f_i)_{i\in I}\) is a net from \(Y^X\), \(f\in Y^X\), \((f_i)\) converges continuously to \(f\) (\(f_i\overset{c} {\rightarrow}f\)) iff \(\forall (x,V)\in X\times\underline{U}(f(x)):\exists (U,k) \in \underline{U}(x)\times I: \forall i\geq k: f_i(U)\subseteq V\). (See for instance the reviewer's book [Compactness in general function spaces (1974; Zbl 0291.54012)] or the more recent paper of \textit{R. Bartsch, P. Dencker}, and \textit{H. Poppe} [in `Categorical topology', Proc. L'Aquila Conf., 221-240 (1996; Zbl 0880.54015)]). A net \((f_i)_{i\in I}\) is called to be topologically convergent to \(f\) (\(f_i\overset{t} {\rightarrow}f\)), iff the net \((\Gamma f_i)\) of the graphs of \(f_i\) Hausdorff-converges to \(\Gamma f\) in the hyperspace of \(X\times Y\) [the reviewer, Math. Res. 67, 256-260 (1992; Zbl 0803.54003)]. Let \(X\) be a topological space and \((Y,\mathfrak{U})\) a uniform space; the author defines: (1) A net \((f_i)_{i\in I}\) is said to be equicontinuous iff for every \(x\in X\) and any \(V\in \mathfrak{U}\) there exists a neighborhood \(U\) of \(x\) and \(i_0\in I\) such that \((f_i(x),f_i(y))\in V\) whenever \(i\geq i_0\) and \(y\in U\). But, using the existence of a \(W\in \mathfrak{U}\) with \(W\circ W\subseteq V\) for every \(V\in \mathfrak{U}\) in a uniform structure \(\mathfrak{U}\), we can easily see, that (1) is just a weakening of (0) for such spaces as mentioned in (1), so we get at once: (2) \(f_i\overset{c} {\rightarrow}f\) \(\Rightarrow\) \((f_i)_{i\in I}\) is equicontinuous, and (3) \(f_i\overset{\tau_p} {\rightarrow}f\) (pointwise) and \((f_i)_{i\in I}\) is equicontinuous \(\Rightarrow\) \(f_i\overset{c} {\rightarrow}f\). The author refers to a paper of the reviewer [Fundam. Math. 59, 159-169 (1966; Zbl 0139.40404)], and he uses the notion of rim-compactness defined in this paper. The reviewer proved in this paper: (4) Let \(X\) be a topological space, \(Y\) a Hausdorff space, \((f_i)_{i\in I}\) a net from \(C(X,Y)\) (or \(Y^X\)). Then holds \(f_i\overset{c} {\rightarrow}f\in C(X,Y)\) \(\Rightarrow\) \(f_i\overset{t} {\rightarrow} f\) (theorem 13); (5) Let \(X\) be a locally connected space and \(Y\) a Hausdorff rim-compact space, \((f_i)_{i\in I}\) a net from \(C(X,Y)\), then \(f_i\overset{t} {\rightarrow}f\in C(X,Y)\) \(\Rightarrow\) \(f_i\overset{c} {\rightarrow}f\) (theorem 14). ((4) was first proved by \textit{O. Frink jun.} [Trans. Am. Math. Soc. 51, 569-582 (1942; Zbl 0061.39305)].) All spaces in the author's paper are regular \(T_1\). He states in lemma 1: Let \(X\) be a locally connected space and \(Y\) a rim-compact space; for a net \((f_i)_{i\in I}\) from \(C(X,Y)\) and \(f\in C(X,Y)\) holds: \(f_i\overset{t} {\rightarrow}f\) \(\Rightarrow\) \(f_i\overset{p} {\rightarrow}f\). Thus the author uses the assumptions of the reviewer, also he adapts the main line of reviewer's proof without mentioning (5) and proves a weaker result than in (5), since clearly continuous convergence implies pointwise convergence. In lemma 2 the author states: Let \(X\) be a locally connected space and \((Y,\mathfrak{U})\) a rim-compact uniform space. For a net \((f_i)_{i\in I}\) from \(C(X,Y)\) topological convergence to \(f\in C(X,Y)\) implies equicontinuity. But without using the author's long proof, from (5) we get easily \(f_i\overset{t} {\rightarrow}f\) \(\Rightarrow\) \(f_i\overset{c} {\rightarrow}f\) in the above situation, and so the equicontinuity follows by (2). In lemma 3 the author proves: Let \(X\) be a topological space and \((Y,\mathfrak{U})\) a uniform space. Let \((f_i)_{i\in I}\) be equicontinuous and \(f_i\overset{\tau_p} {\rightarrow}f\in C(X,Y)\); then \(f_i\overset{t} {\rightarrow}f\). This result at once follows from (3) and (4). Compare here also [the reviewer, Real. Anal. Exch. 18, No. 2, 459-464 (1993; Zbl 0780.54015)]. The author uses his lemmas to prove an Ascoli-theorem (theorem 2): Let \(X\) be a locally connected space, \((Y,\mathfrak{U})\) a rim-compact uniform space and \(\Omega\subseteq C(X,Y)\). If any net from \(\Omega\) has a topologically convergent subnet, then \(\overline{\Omega}\) is compact with respect to the compact-open topology on \(C(X,Y)\). In the opinion of the reviewer, the extent of new information in the author's Ascoli-theorem is small. For a (regular \(T_1\)) locally connected space \(X\) and a Hausdorff rim-compact uniform space \(Y\) the author assumes that a set \(\Omega\) is relatively compact with respect to topological convergence: each net from \(\Omega\) has a subnet converging topologically to some \(f\in C(X,Y)\). But by author's assumptions on \(X\) and \(Y\) we get from (4) and (5) at once, that equivalently \(\Omega\) is relatively compact with respect to continuous convergence, and hence \(\Omega\) is relatively compact with respect to the compact-open topology \(\tau_k\), since \(\tau_k\) is splitting [see for instance: Compactness in general function spaces, loc. cit.]. By the assumptions of the theorem then \((C(X,Y),\tau_k)\) is regular and hence \(\overline{\Omega}^{\tau_k}\) is \(\tau_k\)-compact, too, which is the assertion of the theorem.