A note on \(\beta D-D\) (Q1100760)

A subspace Y of a space X is said to be \(C^*\)-embedded if every bounded, continuous, real-valued function from Y extends continuously over X. The authors produce many examples of spaces which are, and others which are not, \(C^*\)-embedded in the space \(U(\kappa)\) of uniform ultrafiltes over an infinite discrete space \(\kappa\) (here, \(U(\kappa)\) is a subspace of the Stone-Čech compactification of the discrete space \(\kappa)\). Given infinite \(\kappa\), one says that Solovay's Lemma for \(\kappa\) (denoted \(SL_{\kappa})\) is valid provided that if \(\omega \leq \lambda <2^{\kappa}\) and if \(\{F_ i\}_{i<\lambda}\) and \(\{G_ j\}_{j<\lambda}\) are families of subsets of \(\kappa\) such that \(| G_ j\setminus \cup_{i\in S}F_ i| =\kappa\) whenever \(| S| <\kappa\), then there is \(B\subseteq \kappa\) such that \(| B\cap F_ i| =\kappa\) and \(| B\cap G_ i| =\kappa\) for all \(i<\kappa\). The authors' main result is Theorem 2.1: If \(\kappa\) is regular and \(SL_{\kappa}\), then for \(\kappa \leq \lambda <2^{\kappa}\) every cellular family of cardinality \(\lambda\) in U(\(\kappa)\) admits a \(C^*\)-embedded selection. (It is not known whether every selection for such a family is \(C^*\)-embedded.) As the authors point out, the result is not new for \(\lambda =\kappa\) - indeed for \(\kappa\) regular every union of \(\kappa\)-many clopen subsets of \(U(\kappa)\) is \(C^*\)-embedded. Now denote by \(H(\kappa)\) (reviewer's notation) the statement that there is in U(\(\kappa)\) a cellular family of cardinality \(\kappa^+\) with no \(C^*\)-embedded selection. It is known that H(\(\omega)\) is consistent with Lusin's hypothesis \(2^{\omega}=2^{(\omega_ 1)}\) - in fact it is consistent with Lusin's hypothesis that there is in U(\(\omega)\) a P- point with character \(\omega_ 1\). In {\S} 3 the authors prove \(H(\omega_ 1)\) assuming \(2^{(\omega_ 1)}=\omega_ 2\), and they investigate the consistency of \(H(\omega_ 1)\) relative to \(2^{(\omega_ 1)}=2^{(\omega_ 2)}\).