An approximation topology using countable elementary submodels

MaRDI QIDQ6919289zbMATH OpenFDO

Publication date 8 October 2025

Published in Topology Proceedings (Search for Journal in Brave)

Full work available at URL https://topology.nipissingu.ca/tp/restricted/v66/tp66008.pdf

\(\Sigma\)-product elementary submodel metrization theorem

Models with special properties (saturated, rigid, etc.) (03C50) Basic constructions in general topology (54B99) Noncompact covering properties (paracompact, Lindelöf, etc.) (54D20) (p)-spaces, (M)-spaces, (sigma)-spaces, etc. (54E18) Metric spaces, metrizability (54E35)

Review text

It was clearly shown in [\textit{A. Dow}, Topology Proc. 13, No.1, 17--72 (1988; Zbl 0696.03024)] that elementary submodels can be a very poweful tool in topology. In this article, by applying countable elementary submodels, several known topological theorems are given new alternative proofs.\N\NFor a topological space \(\langle X, TOP(X)\rangle\), there exists an infinite regular cardinal \(\kappa\) such that \(\langle X, TOP(X)\rangle \in H(\kappa)\) where \(H(\kappa)\) is the set of all sets whose transitive closure has size less than \(\kappa\). By the downward Löwenheim-Skolem theorem, there exists a countable elementary submodel \(M\) of \(H(\kappa)\) such that \(X\in M\) and \(TOP(X)\in M\). Then \(\tau_M\) is the topology on \(X\) generated by the union of all pseudometrizable topologies on \(X\) that are coarser than \(TOP(X)\) and belong to \(M\). The topology \(\tau_M\) is pseudometrizable. For \(x\in X\), the pair \(\langle M, x\rangle\) is called a nice pair if there exists a neighborhood base \(\mathcal{B}\) at \(x\) in \(\langle X, \tau_M\rangle\) such that \(\mathcal{B}\subseteq M\). For \(x\in X\), topologies \(\tau_1\) and \(\tau_2\) on \(X\) are called equivalent at \(x\) if, for every \(\tau_1\)-neighborhood \(U\) of \(x\), there exists a \(\tau_2\)-neighborhood \(V\) of \(x\) with \(V\subseteq U\), and, for every \(\tau_2\)-neighborhood \(V\) of \(x\), there exists a \(\tau_1\)-neighborhood \(U\) of \(x\) such that \(U\subseteq V\).\N\NA crucial role in this paper is played by the following lemma. Given a topological space \(\langle X, TOP(X)\rangle\), there exist a pseudometrizable topology \(\mathcal{T}_a\) on \(X\) with \(\mathcal{T}_a\subseteq TOP(X)\) and a family \(\{M_x: x\in X\}\) of countable elementary submodels containing everything relevant, such that, for each \(x\in X\), the pair \(\langle M_x, x\rangle\) is a nice pair, and the topologies \(\tau_{M_x}\) and \(\mathcal{T}_a\) are equivalent at \(x\). Unfortunately the proof of the lemma in the article has a gap. Namely, it is highly unclear if the topology \(\mathcal{T}_a\) defined in the proof is pseudometrizable. However, the author has informed the reviewer that the proof of the lemma can be cured as follows. In the original proof, the author constructs a tree \(S\) of height \(\omega\), families \(\{M_s: s\in S\}\) and \(\{M_x: x\in X\}\) of countable elementary submodels, and a family \(\{\mathcal{V}_n: n\in\omega\}\) of locally finite cozero-covers of \(\langle X, TOP(X)\rangle\); the families satisfy several special conditions. In the original proof, \(\mathcal{T}_a\) is the topology on \(X\) generated by \(\bigcup_{n\in\omega}\mathcal{V}_n\). The author cannot check if this topology is pseudometrizable or it may fail to be pseudometrizable. Fortunately, in a letter to the reviewer, the author has suggested the following changes in the proof. Choose a family \(\{M_n: n\in\omega\}\) of countable elementary submodels such that, for every \(n\in\omega\), \(\mathcal{V}_{n+1}\) and \(M_n\) are members of \(M_{n+1}\), and \(M_0=M_X\) is defined in much the same way, as in the original proof of the lemma. The countable elementary submodels are supposed to contain everything relevant. Let \(\mathcal{T}\) be the topology generated by \(\bigcup_{n\in\omega}\mathcal{V}_n\), and, for \(M=\bigcup_{n\in\omega}M_n\), let \(\mathcal{T}_a=\tau_M\). Then \(\mathcal{T}\subseteq \mathcal{T}_a\). Now, the pseudometrizability of \(\mathcal{T}_a\) follows from the fact that \(\tau_M\) is pseudometrizable. It is relatively easy to prove that, for each \(x\in X\), the topologies \(\mathcal{T}\) and \(\tau_{M_x}\) are equivalent, which implies that also \(\tau_{M_x}\) and \(\tau_M=\mathcal{T}_a\) are equivalent. It would be good if a corrected, detailed proof of the lemma could be prepared for publication.\N\NUsing the lemma, the author shows alternative proofs in \(ZFC\) of the following well-known theorems: every compact Hausdorff space with a point-countable base is metrizable; a Hausdorff space \(X\) is metrizable if and only if it is a paracompact \(\beta\)-space with a \(\delta\theta\)-base; a compact Hausdorff space with a \(G_{\delta}\)-diagonal is metrizable; a Hausdorff space is metrizable if and only if it is a submetrizable paracompact \(p\)-space; a Hausdorff space is metrizable if and only if it is a paracompact \(p\)-space with a point-countable \(T_1\)-separating open cover; a \(\Sigma\)-product of Hausdorff, paracompact \(\sigma\)-spaces is normal if it has countable tightness; a \(\Sigma\)-product of Hausdorff semi-stratifiable spaces, each finite subproduct of which is paracompact, is collectionwise normal if it is normal; if \(\langle X, TOP(X)\rangle\) is a Hausdorff paracompact \(\sigma\)-space, and \(\mathcal{U}\) is an interior-preserving open family in \(\langle X, TOP(X)\rangle\), then there exists a metrizable topology \(\tau\) on \(X\) such that \(\tau\subseteq TOP(X)\) and \(\mathcal{U}\) is an interior-preserving open family in \(\langle X, \tau\rangle\). Interesting properties of the classes of spaces considered here in terms of elementary submodels are shown, and useful comments on them are given. However, readers may argue that certain parts of the article lack clarity.

Cites work

This page was built for publication: An approximation topology using countable elementary submodels

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6919289)