Groupwise density and related cardinals (Q918546): Difference between revisions

A family \({\mathcal A}\) of subsets of \(\omega\) is called groupwise dense iff it is closed under almost containment (X\(\in {\mathcal A}\) and Y-X finite implies \(Y\in {\mathcal A})\) and picks up large pieces of partitions by finite sets (if P is a partition of \(\omega\) into finite sets, then some infinite union of elements of P is an element of \({\mathcal A})\). The cardinal number \({\mathfrak g}\) is defined as the smallest cardinal \(\kappa\) so that there are \(\kappa\) many groupwise dense families with no common element. \({\mathfrak g}\) was introduced by the author and LaFlamme as an encapsulation of Shelah's combinatorics relating to NCF (near coherence of filters) and various special types of points in \(\omega^*\). In particular, the assumption \({\mathfrak u}<{\mathfrak g}\) (here \({\mathfrak u}\) is the smallest size of a base for an ultrafilter) suffices to get Shelah's conclusions. Further exploration of the inequality \({\mathfrak u}<{\mathfrak g}\) is continued in papers of the author, LaFlamme and Just, and is continued here, where it is shown that if \({\mathfrak u}<{\mathfrak g}\) then several well-known invariants collapse: \({\mathfrak b}={\mathfrak u}\), and \({\mathfrak d}={\mathfrak g}=c.\) The other main theorems are about \({\mathfrak g}\) itself, and basically say that families (of functions, of sets) of size \(<{\mathfrak g}\) are small. Theorem. If \({\mathcal F}\) is a family of non-decreasing functions in \(^{\omega}\omega\), \(| {\mathcal F}| <{\mathfrak g}\), and \({\mathcal A}\) is an undominated family of non-decreasing functions in \(^{\omega}\omega\), then \({\mathcal F}\) is somewhere dominated by \({\mathcal A}\), i.e., there is some infinite subset A of \(\omega\) and some g in \({\mathcal A}\) so that \(\{\) \(f| A:\) \(f\in {\mathcal F}\}\) is dominated mod finite by \(g| A.\) As a corollary, any two unbounded families of non-decreasing functions in \(^{\omega}\omega\) of size \(<{\mathfrak g}\) somewhere dominate each other. Theorem. If \({\mathcal F}\) is a filter generated by fewer than \({\mathfrak g}\) sets, and \({\mathcal A}\) is a filter which is not feeble, then \({\mathcal F}\leq {\mathcal A}\). Here ``\({\mathcal F}\leq {\mathcal A}''\) means that there is a finite- to-one function f so that f[\({\mathcal F}]\) is a subset of f[\({\mathcal A}].\) As a corollary, any two non-feeble filters generated by fewer than \({\mathfrak g}\) sets dominate each other under the relation \(\leq.\) Notice that (trivially) any unbounded family of non-decreasing functions in \(^{\omega}\omega\) of size \(<{\mathfrak b}\) is bounded, and (not trivially) any filter generated by fewer than \({\mathfrak b}\) sets is feeble, so these results are only of interest when \({\mathfrak b}<{\mathfrak g}\).