Small models, large cardinals, and induced ideals (Q2216033)

Large cardinal properties below the level of a measurable cardinal are often characterized in terms of elementary embeddings of set-sized structures (as opposed to elementary embeddings of the universe \(V\) for stronger notions). Alternatively, these elementary embeddings typically arise as ultrapowers of the domain model \(M\) via some \(M\)-ultrafilter \(U\), and much of the large cardinal property is reflected in the properties of \(U\). In the present paper, the authors set up a general framework to study a variety of these smaller large cardinals. Specifically, they attach to many large cardinal properties \(\Phi(\kappa)\) a property \(\Psi(M,U)\) of models \(M\) and \(M\)-ultrafilters \(U\), so that \(\Phi(\kappa)\) holds if and only if many \emph{nice} models \(M\) have an \(M\)-ultrafilter on \(\kappa\) satisfying \(\Psi(M,U)\). ``Nice'' might mean a number of things depending on the context (but often means that \(M\preceq H_\theta\) for some large \(\theta\), or that \(M\) is transitive of size \(\kappa\)), and ``many'' typically means stationarily many. By varying the exact scheme and combinations of properties including amenability, completeness, and normality for \(U\), the authors are able to characterize inaccessible, weakly compact, ineffable, and different types of Ramsey-like cardinals. The authors also associate natural ideals to their formulas \(\Psi(M,U)\). Depending on the exact notion studied, the ideals \(I^{<\kappa}_\Psi\), \(I^\kappa_\Psi\), or \(I^\kappa_{\prec\Psi}\) contain those sets \(A\subseteq \kappa\) which, for a club of nice models \(M\), do not appear in \emph{any} \(M\)-ultrafilter \(U\) satisfying \(\Psi(M,U)\). As a simple example at the level of measurability, \(I^\kappa_{\prec\text{ms}}\) is the ideal of subsets of \(\kappa\) that do not lie in any normal measure on \(\kappa\). For other large cardinal notions, the ideals are shown to correspond to ideals that were previously defined from more combinatorial characterizations of the large cardinal in questions (such as the weakly compact ideal defined by Lévy, or the ineffable ideal defined by Baumgartner). Importantly, the authors manage to show that inclusions between these ideals correspond strongly to direct implications between the large cardinal properties, and that a large cardinal property \(\Phi_1\) being consistency-wise stronger than \(\Phi_2\) can be read off the ideal associated to \(\Phi_1\) concentrating on the set of non-\(\Phi_2\) cardinals. The paper concludes with an interesting study of the ideal \(I^\kappa_{\prec\text{ms}}\) of sets not included in any normal measure on \(\kappa\) and the quotient of \(\mathcal{P}(\kappa)\) by this ideal. This quotient is atomic in Kunen's model \(L[U]\), since in that model the ideal is prime (it is the complement of the unique normal measure on \(\kappa\)). The authors show that the same conclusion holds in every model in which the Mitchell order on \(\kappa\) has no antichains of size \(\kappa^+\). On the other hand, they also show that it is possible to force over a model with a measurable cardinal \(\kappa\) to make the quotient \(\mathcal{P}/I^\kappa_{\prec\text{ms}}\) atomless. The authors also pose an interesting open question, whether \(I^\kappa_{\prec\text{ms}}\) is necessarily a precipitous ideal when \(\kappa\) is measurable. The paper is very well written, with a careful account of elementary embeddings between nontransitive (or illfounded) models of set theory. The proofs are worked out in reasonable detail and should serve as a good overview of the theory of smallish large cardinals.