Quasi-metrizability of products in ZF and equivalences of CUT(fin) (Q1646538)

In the paper under review, the author studies (quasi)-metrization theorems for products in ZF (Zermelo-Fraenkel set theory without the axiom of choice (AC)), and also provides sufficient conditions for a set \(J\) to be such that the Cantor cube \(2^{J}\) (\(2=\{0,1\}\) has the discrete topology and the product \(2^{J}\) is endowed with the product topology) is both metrizable and non-compact (Sections 2 and 3). A \textit{quasi-metric} on a set \(X\) is a function \(d:X\times X\rightarrow [0,+\infty)\) such that for all \(x,y,z\in X\), the following two conditions are satisfied: (i) \(d(x,y)=0\) \(\Leftrightarrow\) \(x=y\); (ii) \(d(x,y)\leq d(x,z)+d(z,y)\). A topological space \((X,\tau)\) is quasi-metrizable if there exists a quasi metric \(d\) on \(X\) such that \(\tau\) is equal to the topology induced by \(d\) (i.e., \(\tau\) is equal to the topology whose basic open sets are the (open) balls \(B_{d}(x,r)\) \(=\) \(\{y\in X\) : \(d(x,y)\) \(<\) \(r\)\(\}\) where \(x\in X\) and \(r>0\)). With regard to (quasi)-metrization theorems for products (Section 2), the author proves the following interesting main results: \begin{itemize} \item[1.] If \(X\) is a (quasi)-metrizable space which has at least two points, then for a set \(J\), the following are equivalent: (i) \(X^{J}\) is (quasi)-metrizable; (ii) \(X^{J}\) is first-countable; (iii) \(J\) is a countable union of finite sets. (Theorem 2.2.) \item[2.] The following are equivalent: (i) The negation of CC(fin) (where CC(fin) is the axiom of choice for countable families of non-empty finite sets, and which is equivalent to CUT(fin), i.e., ``The union of a countable family of finite sets is countable''); (ii) There exists a set \(J\) such that the Cantor cube \(2^{J}\) is metrizable and not second-countable; (iii) There exists a set \(J\) such that, for every quasi-metrizable space \(X\) consisting of at least two points, the product \(X^{J}\) is quasi-metrizable and not second-countable. (Proposition 2.3.) \item[3.] It is independent of ZF that \(\mathbb{N}^{J}\) is not normal when \(J\) is uncountable. (Corollary 2.5.) \item[4.] For every set \(X\), the following are equivalent: (i) \(\tau_{\mathrm{fin}}(X)\) (the co-finite topology on \(X\)) is non-Archimedeanly quasi-metrizable (a quasi metric \(d\) on \(X\) is non-Archimedean if for all \(x,y,z\in X\), \(d(x,y)\leq\max\{d(x,z),d(z,y)\}\)); (ii) \(\tau_{\mathrm{fin}}(X)\) is quasi-metrizable; (iii) \(\tau_{\mathrm{fin}}(X)\) is first-countable; (iv) \(X\) is a countable union of finite sets. (Theorem 2.7.) \item[5.] CUT(fin) is equivalent to each of ``If \(J\neq\emptyset\) and \(|X|\geq 2\), then the product \(X^{J}\), where \(X\) has the co-finite topology, is quasi-metrizable if and only if both \(J\) and \(X\) are countable'' and ``For every set \(X\), \(\tau_{\mathrm{fin}}(X)\) is (non-Archimedeanly) quasi-metrizable if and only if \(X\) is countable''. (Corollary 2.12.) \item[6.] It is independent of ZF that all countable products of metrizable spaces are quasi-metrizable. (Theorem 2.15.) \end{itemize} A couple of remarks are in order (the first one of those is mentioned in this paper). Firstly, \textit{S. G. da Silva} [Arch. Math. Logic 55, No. 7--8, 867--872 (2016; Zbl 1355.03036)] showed that CUT(fin) is equivalent to each of ``Given any uncountable set \(X\) endowed with the co-finite topology, none of its points has countable character'' and ``For every uncountable set \(X\) with the cofinite topology there is a point which does not have countable character''. Thus, Wajch's results (see items 4 and 5 above) provide new and essential information on topological equivalents of CUT(fin), and strengthen da Silva's result [loc. cit.]. Second, in Theorem 2.15 (see item 6 above), the author essentially proves that ``Countable products of metrizable spaces are quasi-metrizable'' implies \textit{E. K. van Douwen}'s [Proc. Am. Math. Soc. 95, 101--105 (1985; Zbl 0574.03039)] choice principle: ``Every countably infinite family \(\mathcal{A}=\{(A_{i},\preceq_{i}):i\in\omega\}\) of linearly ordered sets isomorphic to \((\mathbb{Z},\leq)\) (\(\leq\) is the usual ordering on \(\mathbb{Z}\)) has a choice function''. However, this result can be enhanced by showing that the above topological statement implies the following weak choice principle: ``For every countably infinite family \(\mathcal{A}\) of countably infinite sets there is a function \(F\) with domain \(\mathcal{A}\) such that for every \(x\in\mathcal{A}\), \(F(x)\) is a non-empty finite subset of \(x\)'' (\(F\) is called a multiple choice function for \(\mathcal{A}\)), which is formally stronger than van Douwen's choice principle [loc. cit.]. (Let \(\mathcal{A}=\{A_{n}:n\in\omega\}\) be a countably infinite family of countably infinite sets, which we assume -- without loss of generality -- to be disjoint. Let \((\infty_{n})_{n\in\omega}\) be an injective sequence whose range is disjoint from \(\bigcup\mathcal{A}\), and also let \(X_{n}=A_{n}\cup\{\infty_{n}\}\) (\(n\in\omega\)) be endowed with the topology of the one-point compactification of \(A_{n}\) with the discrete topology. Then \(X_{n}\) is metrizable for all \(n\in\omega\). If the product space \(X\) of the \(X_{n}\)'s is quasi-metrizable, then we let \(\rho\) be a quasi metric on \(X\) inducing the product topology on \(X\), and we also let \(\mathbf{x}=(\infty_{n})_{n\in\omega}\). Then the function \(F=\{(A_{n},X_{n}\setminus \operatorname{pr}_{n}(B_{\rho}(\mathbf{x},\frac{1}{k_{n}}))):n\in\omega\}\), where \(\operatorname{pr}_{n}\) is the canonical projection of \(X\) onto \(X_{n}\) and \(k_{n}=\min\{k\in\omega\setminus\{0\}: \operatorname{pr}_{n}(B_{\rho}(\mathbf{x},\frac{1}{k}))\subsetneqq X_{n}\}\), is a multiple choice function for \(\mathcal{A}\).) With regard to the part of this paper which studies Cantor cubes in ZF (Section 3), the author proves the following results: \begin{itemize} \item[(a)] Let \(\{X_{n}:n\in\omega\}\) be a collection of non-empty finite discrete spaces. Put \(X=\prod_{n\in\omega}X_{n}\) and \(J=\bigcup_{n\in\omega}(X_{n}\times\{n\})\). If either \(X\) is non-compact or \(X=\emptyset\), then the Cantor cube \(2^{J}\) is non-compact. (Theorem 3.1.) \item[(b)] Let \(\{X_{n}:n\in\omega\}\) be a collection of non-empty finite sets and let \(J=\bigcup_{n\in\omega}X_{n}\). Suppose that \(\prod_{n\in M}X_{n}=\emptyset\) for each infinite set \(M\subseteq\omega\). Then \(2^{J}\) is non-compact (Corollary 3.2.) \item[(c)] Let \(k\in\omega\setminus\{0,1\}\) and, for each \(n\in\omega\), let \(X_{n}\) be a set which has at most \(k\) elements. Suppose that the set \(J=\bigcup_{n\in\omega}X_{n}\) is uncountable. Then \(2^{J}\) is non-compact. (Proposition 3.3.) \item[(d)] It is independent of ZF that every metrizable Cantor cube is compact. In Cohen's second model, there exist Cantor cubes that are simultaneously metrizable and non-compact. (Corollary 3.4.) \end{itemize} A few remarks follow. The first part of (the conclusion of) Theorem 3.1 (see item (a) above), namely that if \(X\) is non-compact then \(2^{J}\) is non-compact, follows immediately from a more general result established by \textit{K. Keremedis} [Math. Log. Q. 58, No. 3, 130--138 (2012; Zbl 1251.03054)]. This result by Keremedis (Proposition 3.2 in his paper) states the following: ``Let \((X_{i})_{i\in I}\) be a disjoint family of discrete spaces and let \(X\) be their product space. If \(Y=\bigcup\{X_{i}:i\in I\}\), and \(Z=\{f\in 2^{Y}:\forall i\in I,|f^{-1}(\{1\})\cap X_{i}|=1\}\) carries the subspace topology, then the function \(f:X\rightarrow Z\) defined by \(f(x)=\chi_{\{x(i):i\in I\}}\) (the characteristic function of \(\{x(i):i\in I\}\)) is a homeomorphism. In addition, if each \(X_{i}\) is finite then \(Z\) is a closed subspace of \(2^{Y}\).'' The second part of (the conclusion of) Theorem 3.1 (i.e., if \(X=\emptyset\) then \(2^{J}\) is non-compact), as well as Proposition 3.3 (see item (c) above), are special cases of a more general and stronger result established by \textit{K. Keremedis} and the reviewer [J. Symb. Log. 75, No. 3, 996--1006 (2010; Zbl 1208.03051)], which states the following: ``Let \(I\) be a set. Each of the following statements implies the one beneath it: (1) \(2^{I}\) is compact; (2) Every disjoint family of non-empty finite subsets of \(I\) has a choice function; (3) For every positive integer \(n\) and for every ordinal number \(\alpha\), any \(|\alpha|\)-sized family of non-empty subsets of \(I\) each having at most \(n\) elements, has a well-orderable union.'' (see Theorem 4 in the above paper). The paper concludes with Problem 3.5 (``Find a model \(M\) of ZF such that CC(fin) fails in \(M\) and CC(\(\leq n\)) holds in \(M\) for each \(n\in\omega\)'', where CC(\(\leq n\)) states that every countable family of non-empty sets each having at most \(n\) elements has a choice function) and the following two questions: Question 3.6 (``Is there a model of ZF in which a Cantor cube can be simultaneously metrizable, not second-countable and compact?'') and Question 3.8 (``Suppose that \(M\) is a model of ZF + \(\neg\)CUT(fin). Is there in \(M\) a metrizable Cantor cube which does not have Hausdorff compactifications in \(M\)?''). It should be noted here that Problem 3.5 above has been resolved (in the affirmative) by \textit{A. Levy} [Fundam. Math. 50, 475--483 (1962; Zbl 0134.24805)] for ZFA (Zermelo-Fraenkel set theory with atoms) and transferred to ZF by \textit{D. Pincus} [Individuals in Zermelo-Fraenkel set theory. Cambridge, MA: Harvard University (PhD Thesis) (1969); ``The strength of the Hahn-Banach theorem'', Lect. Notes Math. 396, 203--248 (1973)]. A detailed proof can also be found in \textit{T. J. Jech} [The axiom of choice. Amsterdam: Elsevier (1973; Zbl 0259.02051)] (see Theorem 7.11 therein).