Ramsey ultrafilters and the reaping number - Con(\({\mathfrak r}<{\mathfrak u}\)) (Q2640594)

A set x (\(\subseteq \omega)\) reaps a family \({\mathfrak A}\) \((\subseteq [\omega]^{\omega})\) iff for any \(a\in {\mathfrak A}\), both \(a\cap x\) and a- x are infinite. The reaping number, denoted by \({\mathfrak r}\), is the smallest cardinality of a family \({\mathfrak A}\subseteq [\omega]^{\omega}\) that no subset of \(\omega\) reaps. It is easily seen that \(\aleph_ 1\leq {\mathfrak r}\leq {\mathfrak u}\leq 2^{\aleph_ 0}\), where \({\mathfrak u}\) is the smallest cardinality of a base for a uniform ultrafilter on \(\omega\), and that Martin's axiom implies \({\mathfrak r}={\mathfrak u}=2^{\aleph_ 0}.\) The main accomplishment of this paper is working-out of a general scheme for constructing generic models of ZFC in which \({\mathfrak r}=\aleph_ 1\) and \({\mathfrak u}=\aleph_ 2=2^{\aleph_ 0}\), showing that equality of \({\mathfrak r}\) and \({\mathfrak u}\) is not provable in ZFC. Generic real adding forcing notions \(Q_ I\) associated with uniform ideals I on \(\omega\) are defined, and it is shown that in generic extensions of a model of \(ZFC+GCH\) constructed from countable support iterations of the form \(<P_{\alpha},Q_{I_{\alpha}}|\alpha <\omega_ 2>\), we have: (a) \(2^{\aleph_ 0}=\aleph_ 2\); (b) the family of all ground model elements of \([\omega]^{\omega}\) cannot be reaped (so \({\mathfrak r}=\aleph_ 1)\); and (c) \({\mathfrak u}=\aleph_ 2\), provided the ideals \(I_{\alpha}\) are chosen with standard bookkeeping care to make sure that the generic reals added by \(Q_{I_{\alpha}}\) neutralize all sets that could end up keeping \({\mathfrak u}\) small if left unattended. By working further fine tuning into the general scheme, a generic model in which \({\mathfrak r}=\aleph_ 1\wedge {\mathfrak u}={\mathfrak i}={\mathfrak s}={\mathfrak d}=\aleph_ 2=2^{\aleph_ 0}\) and one in which \({\mathfrak r}={\mathfrak s}={\mathfrak d}=\aleph_ 1\wedge {\mathfrak u}={\mathfrak i}=\aleph_ 2=2^{\aleph_ 0}\) are produced, where \({\mathfrak i},{\mathfrak s}\) and \({\mathfrak d}\) are respectively the smallest cardinalities of a maximal independent family of subsets of \(\omega\), of a splitting family and of a dominating family in \(\omega^{\omega}.\) Investigation of the forcing notions \(Q_ I\), taking up more than twelve pages and conducted over three sections, is the central part of the whole endeavor. Most of the crucial properties of \(Q_ I\) (e.g. each \(Q_ I\) is proper, and forcing with \(Q_ I\) adds a generic real that prevents I from generating a maximal ideal in the extension even if I is maximal in the ground model) are proved more or less along standard lines in a relatively straightforward manner. One property, which plays a key role in ensuring (b) above and asserts that forcing with \(Q_ I\) leaves most Ramsey ultrafilters in the ground model still generating Ramsey ultrafilters in the extension, requires a long and involved process of proof in which a wide range of combinatorial paraphernalia from games using ideals to a structure called word systems to a partition theorem are brought to bear.