The directedness of the Rudin-Keisler order at measurable cardinals
Yair Hayut, Alejandro Poveda
TL;DR
This work identifies a bridge between the directedness of the Rudin-Keisler order on $\kappa$-complete ultrafilters and a Gluing Property (GP) constructed via a non-stationary Prikry-type forcing. By developing the Gluing Poset and a Coding Lemma, the authors produce models where $\kappa$—measurable with $o(\kappa)=\lambda^+$ or $o(\kappa)=\kappa$—exhibits strong compactness-like directedness for $\mathfrak{U}_{\kappa}$ at a much lower consistency strength than a full $\kappa$-compact cardinal. The main theorem shows equiconsistency between the RK-directedness of $\mathfrak{U}_{\kappa}$ and large-cardinal data about $o(\kappa)$, thereby offering the first compactness-type property at the measurable level with relatively weak assumptions. The Coding Lemma yields a complete map of $\kappa$-complete ultrafilters in the gluing extension, enabling precise control of ultrafilter structures and a negative result in Gitik’s model (GP fails). Open problems point to the precise strength of various GP variants and their connections to strong cardinals.
Abstract
The manuscript is concerned with the Rudin-Keisler order of ultrafilters on measurable cardinals. The main theorem proved read as follows: Given regular cardinals $λ\leq κ$, the following theories are equiconsistent modulo ZFC: (1) $κ$ is a measurable cardinal with $o(κ)=λ^+$ (resp. $o(κ)=κ$). (2) The Rudin-Keisler order restricted to the set of $κ$-complete (non-principal) ultrafilters on $κ$ is $λ^+$-directed (resp. $κ^+$-directed). The theorem reported here is proved after bridging the directedness of the RK-order with the $λ$-Gluing Property introduced by the authors in \cite{HP}. Our result provides what seems to be the first example of a compactness-type property at the level of measurable cardinals whose consistency strength is much lower than the existence of a strong cardinal. As part of our analysis we also answer a question of Gitik by showing that the $\aleph_0$-Gluing Property fails in his classical model from ''Changing cofinalities and the nonstationary ideal". As a consequence of this, in Gitik's model the Rudin-Keisler order fails to be $\aleph_1$-directed.
