Tall cardinals in extender models (Q2075271)

The concept of tall cardinal is a natural weakening of the notion of a strong cardinal ``tall is to strong as strongly compact is to supercompact''. \begin{itemize} \item \(\kappa\) is \(\lambda\)-strong when there exists an elementary embedding \(j:V\to M\) with critical point \(\kappa\), \(j(\kappa)>\lambda\) and \(V_\lambda\subseteq M\). We call \(\kappa\) strong, when it is \(\lambda\)-strong for every \(\lambda\). Furthermore, \(\kappa\) is strong up to \(\lambda\) if it is \(\mu\)-strong for every \(\mu<\lambda\). \item The cardinal \(\kappa\) is \(\lambda\)-tall if there is an elementary embedding \(\pi:V\to M\) with critical point \(\kappa\), \(j(\kappa)>\lambda\) and \(M^\kappa\subseteq M\); \(\kappa\) is tall when it is \(\lambda\)-tall for every \(\lambda\). \end{itemize} A \(\lambda\)-strong cardinal is \(\lambda\)-tall, and any strong cardinal is tall. It can be proved that a measurable limit of tall cardinals is tall and that starting with a strong cardinal, a model can be produced where the least measurable cardinal is tall. The result of the paper under review is very easy to state, but very difficult to prove. The authors show that in certain extender models, a cardinal \(\kappa\) is a tall cardinal if and only if it is either a strong cardinal or a measurable limit of strong cardinals. Some illuminating explanations are in order to understand the hypothesis. First of all, extenders \(E\) are generalizations of normal ultrafilters (\(M\)-ultrafilters). They are families of normal ultrafilters with some coherency properties. Any measurable embedding can be coded by a normal ultrafilter, so the embdedding in the definition of \(\alpha\)-tall or \(\alpha\)-strong cardinal can be coded by an extender. These codes allow us to formulate results in ZFC. Extender models \(L[E]\), more generally \(L[\vec{E}]\) for sequences of extenders, are generalizations of Gödel's constructible universe. Some large cardinals can live there, like Ramsey, Erdös, strong, etc. To deal with sequences of extenders, there are different ways to organize the sequence. The most used method is Jensen's \(\lambda\)-indexing. A crucial issue with these models is that of iterability in order to iterate the construction of ultrapowers. The authors assume that \(L[E]\) is self-iterable. In some sense, it means that \(L[E]\) itself ``knows'' that it is iterable. In order to iterate ultrapowers, it is necessary to work with iteration trees, which can have several ``problematic'' branches. Here, another important assumption appears to avoid this obstacle: there is no inner model with a Woodin cardinal. Now, we can establish the main result. Suppose there is no inner model with a Woodin cardinal and that \(V=L[\vec{E}]\) which is iterable. The following assertions are equivalent for a cardinal \(\kappa\): \(\kappa\) is a tall cardinal if and only if \(\kappa\) is either a strong cardinal or a measurable limit of strong cardinals. The paper is very well written, but difficult to follow because of the strong background needed.