Stone duality for first order logic (Q1105598)

At the beginning of the introduction the author says: ``The most interesting phenomena in model theory are conclusions concerning the syntactical structure of a first order theory drawn from the examination of the models of the theory. With these phenomena in mind, it is natural to ask if it is possible to endow the collection of models of the theory with a natural abstract structure so that from the resulting entity one can fully recover the theory as a syntactical structure. We report here on results intended to constitute a positive answer to this question.'' The positive answer to the same question for propositional logic is well known and the ``natural abstract structure'' is in this case just the topology, the space of models is just the Stone space of the Boolean algebra corresponding to the theory, and the positive answer follows from the Stone duality theorem. The author's answer in the first order case is a very nontrivial analogue of this. Instead of the space of models he considers the category with ultraproducts whose objects are models and morphisms the elementary embeddings. The wonderful theorem 4.1 says that this structure is enough (to recover the theory)ö 1. The category with ultraproducts of models appears as a particular case of the author's notion of an ultracategory, i.e. a category with ultrastructure; an ultrastructure on a category S is an abstract collection of functors U: S \(I\to S\) indexed by pairs (I,U), where I is a set and U an ultrafilter on I, together with certain morphisms between these functors. The reason is that the usual ultraproducts cannot be constructed categorically within the category of models: the construction needs direct products and the projections are not elementary embeddings. 2. We have a list of correspondences (see the paper for detailed information). 3. The categories of pretoposes and ultracategories are actually 2- categories and so the author actually works in the 3-category theory. 4. The results of this paper are closely related to the results of the following papers by the author: ``Stone duality for first order logic'' [Logic Colloqu. '81, Proc. Herbrand Symp. Marseille 1981, Stud. Logic Found. Math., 217-232 (1982; Zbl 0522.03006)], ``A Stone-type representation theory for first order logic'' [Mathematical applications of category theory, Proc. Spec. Sess. 89th Annu. Meet. Am. Math. Soc., Denver/Colo. 1983, Contemp. Math. 30, 175-243 (1984; Zbl 0561.03034)], ``Strong conceptual completeness for first order logic'' [J. Symb. Logic (to appear)] and together with it they form an important part of modern categorical logic - let us call it ``Makkai duality theory''.