A Dedekind finite Borel set (Q627425): Difference between revisions

A set \(D\) is Dedekind finite if there does not exist a one-to-one mapping of \(\omega\) (the set of all natural numbers) into \(D\). (If one assumes the axiom of choice (AC), then Dedekind finite is equivalent to the classical notion of finite; in fact, the latter two notions are equivalent under the countable axiom of choice, i.e., every countable family of non-empty sets has a choice function). In this paper, the author brilliantly proves (Theorem 1.4) that the existence of a Dedekind finite Borel set is relatively consistent with ZF, i.e., Zermelo-Fraenkel set theory minus AC. In order to achieve his goal, the author starts with a countable transitive model \(M\) of ZF in which there exists a Dedekind finite set \(D\subseteq 2^{\omega}\) (e.g., the basic Cohen model). Then, using the notion of almost-disjoint sets forcing of R. Solovay, he defines a suitable poset \((\mathbb{P},\leq)\in M\) of forcing conditions (\(\mathbb{P}\) is the direct sum of the posets \(\mathbb{P}(D_n)\), \(n\in\omega\), where \(D_n\) consists of all functions \(f\in 2^{\omega}\) for which there exists \(d\in D\) such that, \(\forall k<n\), \(f(k)=d(k)\) and \(\exists l>n\) \(\forall k>l\) \(f(k)=d(k)\), and \(\mathbb{P}(D_n)\) is the almost-disjoint sets poset with respect to \(D_n\)) which is \(\sigma\)-centered in \(M\), i.e., \(\mathbb{P}=\bigcup\limits_{n\in\omega}\mathbb{P}_{n}\) such that, for all \(n\in\omega\), any finite number of conditions in \(\mathbb{P}_{n}\) are compatible, that is they have a common lower bound in \(\mathbb{P}\). In this way, the author assures that, if \(G\) is any \(\mathbb{P}\)-generic set over \(M\), then the Dedekind finiteness property of \(D\) is preserved in the extension model \(M[G]\) (this is the result of Lemma 4.2). He then goes on defining a suitable group \(\mathcal G\) of automorphisms of \(\mathbb{P}\) and a normal filter \(\mathcal{F}\) of subgroups of \(\mathcal{G}\) which determine a symmetric model \(\mathcal{N}\) such that \(M\subseteq\mathcal{N}\subseteq M[G]\), and in which he proves that \(D\) is (of course still Dedekind finite due to the above facts and) an \(F_{\sigma\delta}\)-set, i.e., a countable intersection of \(F_{\sigma}\)-sets. The author also shows the following couple of results: {\parindent6.5mm \begin{itemize}\item[(1)] If \(A\subseteq 2^{\omega}\) is a \(G_{\delta\sigma}\)-set (i.e., a countable union of \(G_{\delta}\)-sets), then \(A\) is countable or contains a perfect set (i.e., a set homeomorphic to \(2^{\omega}\)). This is Theorem 1.2. \item[(2)] If \(2^{\omega}\) is a countable union of countable sets (this is relatively consistent with ZF, as was shown by \textit{S. Feferman} and \textit{A. Lévy} [``Independence results in set theory by Cohen's method'', Notices Am. Math. Soc. 10, 593 (1963)]), then there exists an \(F_{\sigma\delta}\)-set \(B\subseteq 2^{\omega}\) which is uncountable but contains no perfect subset. This is Theorem 1.3. In addition, the author points out (Remark 3.5) that under the above assumption on \(2^{\omega}\) there is no onto function \(f:2^{\omega}\rightarrow F_{\sigma\delta}\), answering in this way a related question by Alessandro Andretta. \end{itemize}} In his endeavour, the author also establishes the next two very nice results without invoking any form of choice, namely, ``the countable union of countable closed subsets of \(2^{\omega}\) is countable'' (Corollary 2.4) and ``the countable union of countable \(G_{\delta}\)-subsets of \(2^{\omega}\) is countable'' (Corollary 2.6). A couple of problems left open in the paper are the following: {\parindent5mm \begin{itemize}\item[(i)] Is it possible to have an infinite Borel \(\Delta_{5}\)-set? (A set \(D\) is \(\Delta_{5}\) if there does not exist an onto map \(f:D\rightarrow D\cup\{\infty\}\) where \(\infty\notin D\). The notion of a \(\Delta_{5}\)-set is due to \textit{J. Truss} [``Classes of Dedekind finite cardinals'', Fundam. Math. 84, 187--208 (1974; Zbl 0292.02049)].) \item[(ii)] Is it possible to have an uncountable Borel set \(D\subseteq 2^{\omega}\) such that every subset is countable or co-countable? \end{itemize}}