Characterization of model Mirimanov-von Neumann cumulative sets (Q2462036)

In 1917 Mirimanoff used the power set and union operations together with transfinite induction to define the proper class \(V=\bigcup (V_\alpha\mid \alpha\in On)\), the cumulative hierarchy of sets, where \(V_0=\emptyset, V_{\alpha+1}=\mathcal P(V_\alpha)\) and for \(\lambda\) a limit ordinal \(V_\lambda=\bigcup(V_\alpha\mid \alpha<\lambda)\). Under ZF's Axiom of Regularity every set is a member of \(V\). Zermelo first took up the question of the existence of natural models of ZF, meaning models of the form \((V_\alpha, \in)\). If \(\kappa\) is inaccessible, then \((V_\kappa, \in)\) is a natural model of ZF. Must standard models \((U, \in)\) where \(U\) is a set be of this form? A transitive set that is closed under the power set operation is called supertransitive. Under the assumption of a strong version (second order) of the Axiom of Replacement the answer is: the supertransitive set \(U\) is a standard model of ZF if and only if \(U=V_\kappa\) with \(\kappa\) an inaccessible cardinal. For the NBG (von Neumann-Bernays-Gödel) class theory there is no need to strengthen the replacement axiom and one gets: (Zermelo-Shepherdson) the supertransitive set \(U\) is a standard model of NGB if and only if \(U=V_{\kappa+1}\) with \(\kappa\) an inaccessible cardinal. The authors pursue investigations along these lines. Drawing from the work of category theorists, they call a set \(U\) universal if \(U\) is infinite, supertransitive (and closed under some basic set-theoretic operations, e.g. pairing) and satisfies the following version of (strong) replacement: \((\forall f)(x\in U \wedge f\in U^x\rightarrow \operatorname{rng} f\in U)\). Drawing from work of Tarski on inaccessibles, they call a set \(U\) a Tarski set if \(U\) is supertransitive and satisfies: (\(x\subset U\wedge (\forall f)(f\in U^x\rightarrow \operatorname{rng}f \not=U))\rightarrow x\in U\). The authors prove that every uncountable Tarski set is universal. They thus obtain the following characterization of natural models of NBG. The following are equivalent: (1) \(U=V_\kappa\) for \(\kappa\) inaccessible; (2) \(\mathcal P(U)\) is a supertransitive standard model of NBG; (3) \(U\) is universal; (4) \(U\) is an uncountable Tarski set. The paper contains additional results on universal sets and Tarski sets and their relationship to inaccessible cardinals. A theorem of Montague and Vaught shows that for \(\kappa\) inaccessible there exists an \(\alpha<\kappa\) with \(\text{cf}(\alpha)=\omega\) such that \((V_\alpha,\in)\) is a natural model of ZF (with the usual Axiom Schema of Replacement). To obtain a characterization for natural models of ZF simlar to the one for NBG, the authors introduce the concepts of scheme-inaccessible cardinal, scheme-universal set and scheme Tarski set. \{Reviewer's remarks: The authors include a lot of basic set theory in order to make their work more or less self contained. There are a few minor translation problems. The Axiom of Extensionality becomes the ``volume axiom'' and the Axiom Schema of Separation becomes the ``isolation axiom scheme''. The content of the Montague-Vaught Theorem seems to be misstated, implying that \(\alpha\) is inaccessible.\(\}\)