Infima of quasi-uniform anti-atoms (Q2433694)

This intriguing paper continues the authors' study, initiated in [Topology Appl. 153, No. 16, 3140--3156 (2006; Zbl 1119.54018)], of the lattice \((q(X),\subseteq)\) of all quasi-uniformities on a nonempty set \(X\). While there are results concerning the way the lattice of uniform ties embeds in \((q(X),\subseteq)\) and results concerning the filters \(\text{pr}_1({\mathcal G})\) and \(\text{pr}_2({\mathcal G})\) of the paper just mentioned, the present paper centers on two dual questions: which members of \(q(X)\) are the supremum of a collection of atoms of \((q(X),\subseteq)\) and which members are the infimum of a collection of anti-atoms of \((q(X),\subseteq)\)? The authors dispatch the first question: A quasi-uniformity on a set \(X\) is the supremum of a collection of atoms of \((q(X),\subseteq)\) if and only if it is both totally bounded and transitive. Let \({\mathcal C}\) denote the class of quasi-uniformities that are the infimum of a collection of anti-atoms of \((q(X),\subseteq)\) for some set \(X\). The authors show that a quasi-uniformity \({\mathcal U}\) belongs to \({\mathcal C}\) if and only if for each quasi-pseudometric \(d\) on \(X\) such that \({\mathcal U}_d\nsubseteq{\mathcal U}\) there exists a quasi-pseudometric \(d'\) on \(X\) such that \({\mathcal U}_{d'}\vee{\mathcal U}\) is not the discrete uniformity whereas \({\mathcal U}_{d'}\vee{\mathcal U}_d\) is the discrete uniformity. This characterization may yet prove key, but the authors state that they do not know of a quasi-uniformity that does not belong to \({\mathcal C}\). Here then are some of the authors' partial descriptions of \({\mathcal C}\). Any quasi-uniformity with a countable (or even linearly ordered) base belongs to \({\mathcal C}\). Any proximally fine (in particular any fine) quasi-uniformity belongs to \({\mathcal C}\). If \({\mathcal U}\) is a quasi-uniformity on a set \(X\) whose supremum uniformity \({\mathcal U}\vee{\mathcal U}^{-1}\) generates a resolvable topology, \({\mathcal U}\) belongs to \({\mathcal C}\). Reviewer's remarks: The crucial use of Zorn's Lemma and of ultrafilters, together with the fundamental nature of \({\mathcal C}\) suggest that the question whether or not every quasi-uniformity belongs to \({\mathcal C}\) may depend on set theory.