A standard model of Peano arithmetic with no conservative elementary extension (Q998319)

If \(L\) is a first-order language, then PA\((L)\) is the result of adding to the Peano axioms all the instances of the induction schema for \(L\)-formulas. It is well-known that, for countable \(L\), all models of PA\((L)\) have conservative elementary end extensions. In ``A model of Peano arithmetic with no elementary end extension'' [J. Symb. Log. 43, 563--567 (1978; Zbl 0388.03029)], \textit{G. Mills} constructed a nonstandard model of PA\((L)\) for an \(L\) of cardinality \(\aleph_1\), with no elementary end extension. This left open the question whether the standard model could have an expansion without a conservative elementary extension. Enayat gives a very interesting construction of such a model by first translating the question into one about existence of nonprincipal ultrafilters in the powerset of \(\omega\) in models of ZFC\(^-+V=H(\aleph_1)\), and then by applying \(\diamondsuit_{\aleph_1}\) to get such a model without ultrafilters. Shelah's \(\diamondsuit_{\aleph_1}\) elimination technique is then used to get the result in ZFC. The theorem and the technique of its proof yield two other important corollaries. Kirby and Paris showed that if \(I\) is a strong cut in a countable model \(M\models \text{PA}\), then \(M\) has an elementary extension \(N\) inserting new elements right above \(I\) (and none inside \(I\)) and such that every subset of \(I\) coded in \(N\) is already coded in \(M\). Enayat shows in Theorem B that the assumption that \(M\) is countable is essential. In Theorem C he gives an example of a model of PA\((L)\) with no elementary end extension which is different from the one constructed by Mills. Answering a question of Gitman and Hamkins, he also gives an example of an uncountable arithmetically closed Scott set such that the forcing notion associated with it is not proper (it actually collapses \(\aleph_1\)). The paper concludes with a list of attractive open problems.