Michael spaces and ultrafilters (Q6150399): Difference between revisions

The subject of the paper is Michael's spaces (Lindelöf spaces which have a non-Lindelöf product with the Baire space) and ultrafilters. The most important theorem of the paper relates to Michael's ultrafilter -- an ultrafilter \(\mathscr{U}\) is Michael's if for every compact \(K\subseteq \omega^{\omega}\), if \(\operatorname{cof}(K/\mathscr{U})>\omega\), then \(\operatorname{cof}(K/\mathscr{U})\geqslant \operatorname{cof}(\omega^{\omega}/\mathscr{U})\). Theorem 1.2 states what the relationship is between Michael spaces and Michael ultrafilters - if there is a Michael ultrafilters, then there is a Michael space. The main result of the paper is Theorem 2.7, which states that a selective ultrafilter \(\mathscr{U}\) is Michael if and only if \(\operatorname{cof}(\omega^{\omega}/\mathscr{U})\leqslant \operatorname{cin}(\omega^{\omega}/\mathscr{U})\). The implication to the right runs as follows: \(\operatorname{cin}(\omega^{\omega}/\mathscr{U})=\operatorname{cof}(K_0/\mathscr{U})\geqslant \operatorname{cof}(\omega^{\omega}/\mathscr{U})\). In equality, the author takes advantage of the selectivity of the ultrafilter (exactly being a Q-point) and then he applies Proposition 2.4. which says that \(\operatorname{cin}(\omega^{\omega}/\mathscr{U})=\operatorname{cof}(K_0/\mathscr{U})\). In the inequality, the author uses Michael's definition of ultrafilter, the selectivity of ultrafilter (exactly being a P-point) and Proposition 2.1, which gives us: \(\operatorname{cof}(K_0/\mathscr{U})>\omega\). In the implication in the other direction, the author uses the concept of an internally unbounded: Given a filter \(\mathscr{F}\) on \(\omega\), a compact set \[K=\{f\in \omega^{\omega}: \mathop\forall_{n\in \omega} f_{\restriction{n}}\in T \}\subseteq \omega^{\omega}\] (where \(T\subseteq \omega^{<\omega}\) is a tree) is internally unbounded in \(\mathscr{F}\) if for every\(f\in K\) and every \(s\in T\) there is \(g\in K\) extending \(s\) such that \(\{n\in \omega: g(n)\geqslant f(n)\}\in \mathscr{F}\). It also uses a Lemma 1.4 and Proposition 2.6 related to this concept. Lemma 1.4: Given an ultrafilter \(\mathscr{U}\) and a compact set \(K\subseteq \omega^{\omega}\) such that \(\operatorname{cof}(K/\mathscr{U})>\omega\), there is a compact \(K'\subseteq K\) internally unbounded in \(\mathscr{U}\) such that \(\operatorname{cof}(K/\mathscr{U})=\operatorname{cof}(K'/\mathscr{U})\). Proposition 2.6: Assume \(\mathscr{U}\) is a selective ultrafilter, if \[K=\{f\in \omega^{\omega}: \mathop\forall_{n\in \omega} f_{\restriction{n}}\in T \}\subseteq \omega^{\omega}\] is a compact set internally unbounded on \(\mathscr{U}\) then \(\operatorname{cof}(K/\mathscr{U})\geqslant \operatorname{cof}(K_0/\mathscr{U})\). From Lemma 1.4, we know that there exists a compact set \(K'\subseteq K\) such that \(\operatorname{cof}(K/\mathscr{U})=\operatorname{cof}(K'/\mathscr{U})\). Then, the author gets a sequence of inequalities: \(\operatorname{cof}(K'/\mathscr{U})\geqslant \operatorname{cof}(K_0/\mathscr{U})=\operatorname{cin}(\omega^{\omega}/\mathscr{U})\geqslant \operatorname{cof}(\omega^{\omega}/\mathscr{U})\). The first inequality is obtained from Proposition 2.6, and the equality is obtained from Proposition 2.4. Thus, \(\mathscr{U}\) is Michael's ultrafilter. Under the assumption \(\operatorname{cov}(\mathcal{M})=\mathfrak{c}\) the author shows using transfinite induction that any filter with a character smaller than \(\mathfrak{c}\) can be extended to Michael's ultrafilter (Theorem 1.5). At the end of the paper, the author shows that there is a model without Michael's ultrafilters. Reading the paper, you can find 6 interesting open-ended questions.