Thin ultrafilters and the P-hierarchy of ultrafilters (Q2217243)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Thin ultrafilters and the P-hierarchy of ultrafilters
scientific article

    Statements

    Thin ultrafilters and the P-hierarchy of ultrafilters (English)
    0 references
    0 references
    0 references
    29 December 2020
    0 references
    If \(\mathcal I\) is a family of subsets of some set \(X\), an ultrafilter \(u\) on \( \omega\) is an \textit{\(\mathcal I\)-ultrafilter} if, for every function \(f: \omega \to X\), there is \(U\in u\) such that \(f(U) \in I\) [\textit{J. E. Baumgartner}, J. Symb. Log. 60, No. 2, 624--639 (1995; Zbl 0834.04005)]. Many classes of ultrafilters can be characterized in this way. As noticed, e.g., in [\textit{M. Hrušák} and \textit{D. Meza-Alcántara}, Rend. Ist. Mat. Univ. Trieste 44, 503--511 (2012; Zbl 1284.03229)], Baumgartner's notion of an \(\mathcal I\)-ultrafilter can be expressed in terms of the Katětov order among filters [\textit{M. Katětov}, Commentat. Math. Univ. Carol. 9, 173--189 (1968; Zbl 0155.50301)]. If \(u\) is an \(\mathcal I\)-ultrafilter and \(v \leq u\) in the Rudin-Keisler (pre-)order, then \(v\) is an \(\mathcal I\)-ultrafilter. Intuitively, if \(u\) is an \(\mathcal I\)-ultrafilter, then all ``simpler'' ultrafilters are still \(\mathcal I\)-ultrafilters. On the other hand, a linear way to classify the complexity of ultrafilters, the \(P\)-hierarchy, has been proposed in [\textit{A. Starosolski}, J. Symb. Log. 73, No. 4, 1202--1214 (2008; Zbl 1156.03047)] using sequential contours along a cascade. Contours are also known as limits, Frolík sums or tensor products of filters; a cascade is a tree without infinite branches and with a smallest element. The \(P\)-hierarchy consists of \( \omega_1\) disjoint classes \(\mathcal P_ \alpha \); the \(\mathcal P_ 2 \) class contains exactly the \(P\)-points. Under the Continuum Hypothesis, each \(\mathcal P_ \alpha \) is nonempty. The main result of the paper, together with former results, confirms the idea that the above two ways of classifying ultrafilters, the \(\mathcal I\)-ultrafilters and the \(P\)-hierarchy are, in a sense, orthogonal. An infinite subset \(\{ a_n \, \mid n \in \omega \,\}\) of \( \omega\) is \emph{thin} if \(\lim _{n \to \infty} \frac{a_n}{a_{n+1}} =0 \), where the \(a_n\)'s are enumerated in increasing order [\textit{A. Blass} et al., Proc. Am. Math. Soc. 129, No. 11, 3313--3320 (2001; Zbl 0992.28002)]. \(\mathcal T\) denotes the ideal generated by the thin sets. \textit{J. Flašková} [Ultrafilters and small sets. Prague: Charles University in Prague (Doctoral thesis) (2006)] proved under Martin's Axiom for countable posets that there are \(\mathcal T\)-ultrafilters which are not \(P\)-points. The authors generalize Flašková's result to the whole of the \(P\)-hierarchy. Theorem. Assume Martin's Axiom. If \(\gamma\) is an ordinal and \(1 \leq \gamma < \omega _1\), then there exists a \(\mathcal T\)-ultrafilter which belongs to \(P_ \gamma \).
    0 references
    0 references
    0 references
    0 references
    0 references
    P-hierarchy
    0 references
    Martin's axiom
    0 references
    P-points
    0 references
    monotone sequential contour
    0 references
    thin ultrafilter
    0 references
    0 references