Magidor-like forcing and the cofinality of the Galvin number

Shimon Garti · 2025

Read Magidor-like forcing and the cofinality of the Galvin number by Shimon Garti (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-12804.txt.

Introduction

Shimon Garti proves a consistency result about a cardinal number from set theory in a 2025 mathematics study. Starting from a model of the axiom of determinacy. He uses a forcing construction modeled on Magidor's to obtain a universe where the Galvin number, a measure from infinite combinatorics, has countable cofinality. Consistency results of this kind map the borders of what the axioms of set theory allow.

Catalog record

CREATORShimon Garti
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-12804

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Magidor-like forcing and the cofinality of the Galvin number Thanks: Research supported by ISF grant 2320/23. Shimon Garti Abstract . We prove that one can force over a model of AD to obtain countable cofinality of the full Galvin number. Nous prouvons que on peut forcer sur un modèle d’AD pour obtenir cofinalité dénombrable du nombre de Galvin. 2010 Mathematics Subject Classification 03E60, 03E35, 03E55 0. Introduction Let \mathscr{F} be a normal filter over \kappa , where \kappa is strongly regular and uncountable. 1 1 1 A cardinal \kappa is strongly regular if and only if \kappa=\kappa^{<\kappa} . Galvin proved that every family \mathcal{C}=\{C_{\alpha}:\alpha\in\kappa^{+}\}\subseteq\mathscr{F} admits a subfamily \mathcal{D}=\{C_{\alpha_{i}}:i\in\kappa\} so that \bigcap\mathcal{D}\in\mathscr{F} , see [ BHM75 ] . In particular, if 2^{\omega}=\omega_{1} then the club filter of \aleph_{1} satisfies this property. Given \aleph_{0}<\kappa\leq\partial\leq\lambda and a normal filter \mathscr{F} over \kappa let us write {\rm Gal}(\mathscr{F},\partial,\lambda) to denote the statement that every \mathcal{C}\subseteq\mathscr{F} of size \lambda admits a subfamily \mathcal{D}\subseteq\mathcal{C} of size \partial such that \bigcap\mathcal{D}\in\mathscr{F} . Using this terminology, Galvin’s theorem says that if 2^{\omega}=\omega_{1} then {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\aleph_{2}) holds, where \mathscr{D}_{\aleph_{1}} is the club filter of \aleph_{1} . Abraham and Shelah proved in [ AS86 ] that one can force 2^{\omega_{1}}=\lambda along with \neg{\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\lambda) , where \lambda is arbitrarily large. An interesting feature of the proof is that 2^{\omega}=\lambda in this model as well. It was shown in [ Gar17 ] that this fact is not just a coincidence. To wit, if \mu=(2^{\omega})^{+} then {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\mu) holds. From the definition it follows that if \mu^{\prime}\geq\mu then {\rm Gal}(\mathscr{F},\partial,\mu) implies {\rm Gal}(\mathscr{F},\partial,\mu^{\prime}) . Thus, if 2^{\omega}<\lambda then {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\lambda) follows. One concludes that Galvin’s property for clubs of \aleph_{1} depends on the size of 2^{\omega} . This fact is quite surprising, as one could expect that Galvin’s property at \aleph_{1} will be determined by the number of clubs of \aleph_{1} , that is 2^{\omega_{1}} . The fact that 2^{\omega} plays an important rôle here, gives rise to the following definition of \mathfrak{gp} from [ Gar18 ] , which places it in the area of cardinal characteristics of the continuum: Definition 0.1 . The Galvin number. Let \mathfrak{gp} be the minimal \kappa such that every family \mathcal{C}=\{C_{\alpha}:\alpha\in\kappa^{+}\} of clubs of \aleph_{1} contains a subfamily \mathcal{D}=\{C_{\alpha_{i}}:i\in\omega_{1}\} so that \bigcap\mathcal{D}\in\mathscr{D}_{\aleph_{1}} . A natural question is what the possible values of \mathfrak{gp} are. In the model of Abraham and Shelah, one begins with an infinite cardinal \lambda and forces 2^{\omega}=2^{\omega_{1}}=\lambda along with the failure of {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\lambda) . Observe that {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},(2^{\aleph_{1}})^{+}) is trivial, so {\rm Gal}(\mathscr{D}_{\aleph_{1}},\aleph_{1},\lambda^{+}) holds in this setting. Consequently, \mathfrak{gp}=\lambda in this model. This forcing construction applies to every \lambda so that {\rm cf}(\lambda)>\aleph_{1} , hence \mathfrak{gp} can be either regular or singular of cofinality above \aleph_{1} , at every \lambda in the \aleph -scale. This pattern does not immediately work when {\rm cf}(\lambda)\leq\aleph_{1} , simply because 2^{\omega_{1}}=\lambda in the Abraham-Shelah model. However, if {\rm cf}(\lambda)=\omega_{1} then one can force \mathfrak{gp}=\lambda , essentially by the same forcing construction, as will be shown in the last section of this paper. The case of countable cofinality is more interesting. In particular, it was asked in [ Gar18 , Question 4.4] whether {\rm cf}(\mathfrak{gp})=\omega is consistent. Though the question is still open, we know from [ GHHM22 ] that if {\rm cf}(\mathfrak{gp})=\omega is consistent then it has some consistency strength. The purpose of this paper is to build a model of ZF in which the cofinality of the full Galvin number is countable. The full Galvin number \mathfrak{fgp} is a variation of \mathfrak{gp} , adapted to the context of AD . A formal definition will be given in the next section. The rest of the paper is arranged in three sections. In the first section we give some background and in the second section we prove the main result. In the last section we discuss some ZFC results with regard to the cofinality of the Galvin number. Our notation is hopefully standard. We shall employ the Jerusalem forcing notation, that is, we force upwards. We denote the club filter over \kappa by \mathscr{D}_{\kappa} whenever \kappa is regular and uncountable. The collection of \omega -closed unbounded subsets of \kappa is denoted by \mathscr{U}^{\kappa}_{\omega} . In the case of \kappa=\aleph_{1} this is simply the club filter \mathscr{D}_{\aleph_{1}} . We let \Theta be the supremum of \lambda for which there is a surjection \pi:{}^{\omega}\omega\rightarrow\lambda . Finally, a cardinal \kappa is a boldface GCH cardinal iff there is no injection f:\kappa^{+}\rightarrow\mathcal{P}(\kappa) . For background in pcf theory we refer to [ AM10 ] and to Shelah’s monograph [ She94 ] . For background in the combinatorial aspects of AD we suggest the excellent monograph of Kleinberg, [ Kle77 ] . Basic information about polarized relations can be found in [ Wil77 ] . 1. Preliminaries Seemingly, Galvin’s property is connected to the axiom of choice. The assumption \kappa=\kappa^{<\kappa}>\aleph_{0} within the statement of Galvin’s theorem requires some choice, and the proof meets this axiom as well. However, instances of Galvin’s property are provable even without full choice. An initial study of this phenomenon is carried out in [ BGP22 ] . Here we phrase and prove several additional statements. Ahead of our first claim, recall that \binom{\alpha}{\beta}\rightarrow\binom{\gamma}{\delta} denotes the statement that for every coloring c:\alpha\times\beta\rightarrow\{0,1\} one can find A\in[\alpha]^{\gamma},B\in[\beta]^{\delta} so that c\upharpoonright(A\times{B}) is constant. One can replace the ordinals in this relation by some structure, with the natural intended meaning. Thus, if \mathscr{W}\subseteq\mathcal{P}(\lambda) and \mathscr{U}\subseteq\mathcal{P}(\kappa) then \binom{\lambda}{\kappa}\rightarrow\binom{\mathscr{W}}{\mathscr{U}} means that the sets A,B which form the monochromatic product belong to \mathscr{W} and \mathscr{U} respectively. Claim 1.1 . Suppose that \aleph_{0}<\kappa<\lambda and both \kappa and \lambda are measurable. Then {\rm Gal}(\mathscr{U},\lambda,\lambda) holds whenever \mathscr{U} is a normal ultrafilter over \kappa . Proof . Let \mathscr{U} be a normal ultrafilter over \kappa , and let \mathscr{W} be a normal ultrafilter over \lambda . As a first step we prove the combinatorial relation \binom{\lambda}{\kappa}\rightarrow\binom{\mathscr{W}}{\mathscr{U}} . Then we derive the statement of the claim from this relation. Assume, therefore, that c:\lambda\times\kappa\rightarrow{2} is given. For every \beta\in\kappa there is a unique j_{\beta}\in\{0,1\} for which S^{j_{\beta}}_{\beta}=\{\alpha\in\lambda:c(\alpha,\beta)=j_{\beta}\} belongs to \mathscr{W} . It follows that for some fixed j and a set B\in\mathscr{U} one has j_{\beta}=j whenever \beta\in{B} . Define A=\bigcap\{S^{j}_{\beta}:\beta\in{B}\} and notice that A\in\mathscr{W} by \lambda -completeness. Observe that c^{\prime\prime}(A\times{B})=\{j\} , hence we are done proving that \binom{\lambda}{\kappa}\rightarrow\binom{\mathscr{W}}{\mathscr{U}} . Assume now that \{C_{\alpha}:\alpha\in\lambda\}\subseteq\mathscr{U} . Define d:\lambda\times\kappa\rightarrow{2} by letting d(\alpha,\beta)=0 iff \beta\in{C_{\alpha}} . Let A\in\mathscr{W},B\in\mathscr{U} be such that d\upharpoonright(A\times{B}) is constant. Notice that the constant value cannot be 1 . Indeed, choose \alpha\in{A} and any element \beta\in B\cap C_{\alpha} and conclude that \beta\notin C_{\alpha} if d^{\prime\prime}(A\times{B})=\{1\} , contradicting the choice of \beta . Thus d^{\prime\prime}(A\times{B})=\{0\} , and hence B\subseteq\bigcap\{C_{\alpha}:\alpha\in{A}\} . Since B\in\mathscr{U} , the proof is accomplished. \qed_{\ref{clmmeasurables}} Let us indicate that in order to obtain the statement {\rm Gal}(\mathscr{U},\lambda,\lambda) from the relevant polarized relation one has to assume that \kappa , the domain of \mathscr{U} , is measurable, but the assumption that \lambda is also measurable can be relaxed. A moment of perusal shows that if \binom{\lambda}{\kappa}\rightarrow\binom{\lambda}{\mathscr{U}} then {\rm Gal}(\mathscr{U},\lambda,\lambda) obtains. The argument is just the same, and this fact might be helpful. The statement {\rm Gal}(\mathscr{U},\lambda,\lambda) is a strong form of the Galvin property. The above claim

Read the complete text