Elementary cuts in saturated models of Peano arithmetic (Q424574)

Elementary cuts in countable recursively saturated models of PA have been studied by Kotlarski, Smoryński, and others; see for example [\textit{H. Kotlarski}, ``On elementary cuts in recursively saturated models of Peano arithmetic'', Fundam. Math. 120, 205--222 (1984; Zbl 0572.03016)] and [\textit{C. Smoryński}, ``Elementary extensions of recursively saturated models of arithmetic'', Notre Dame J. Formal Logic 22, 193--218 (1981; Zbl 0503.03032)]. Much has been done, and there are still interesting open problems. Schmerl considers analogous questions for saturated models, and comes up with an almost complete set of answers. Let \(\kappa\) be an uncountable cardinal such that the unique \(\kappa\)-dense linear ordering \(\eta_\kappa\) exists. Schmerl defines a more general notion of a boundedly saturated model of cardinality \(\kappa\) and characterizes such models as those whose reduced order type is \(\eta_\kappa\cdot\lambda\), where \(\lambda\leq\kappa\) is regular. Then he shows that for every completion \(T\) of PA, there are exactly \(2^{\aleph_0}\) models having reduced order type \(\eta_\kappa\cdot\omega\) and for every uncountable regular \(\lambda\) there is, up to isomorphism, exactly one model of \(T\) having reduced order type \(\eta_\kappa\cdot\lambda\). The main results concern isomorphism types and first-order theories of pairs \((N,M)\), where \(N\) is a saturated model of PA and \(M\) is an elementary initial segment. An almost complete list of isomorphism types is given, and the rest of the paper is devoted to theories of such pairs. Let \(\kappa\) be the cardinality of \(N\). An elementary cut \(M\) is \textit{balanced} if its upward and downward cofinality in \(N\) is \(\kappa\), otherwise it is \textit{unbalanced}. Schmerl shows that there are \(2^{\aleph_0}\) theories of \((N,M)\) where \(M\) is balanced. If \(M\) is unbalanced, then there are essentially two cases. The situation is completely analyzed if either upward or downward cofinality of \(M\) is \(\aleph_0\), in particular, there are \(2^{\aleph_0}\) theories of \((N,M)\) where \(M\) is short. The most interesting is the case of unbalanced \(M\) with either upward or downward uncountable cofinality. It turns out that there is only one theory of such pairs, and the proof is based on a delicate combinatorial argument involving an Ehrenfeucht-Fraïssé game.