The primal framework. I

John T. Baldwin, Saharon Shelah

Published 1990-01-15Version 1

This the first of a series of articles dealing with abstract classification theory. The apparatus to assign systems of cardinal invariants to models of a first order theory (or determine its impossibility) is developed in [Sh:a]. It is natural to try to extend this theory to classes of models which are described in other ways. Work on the classification theory for nonelementary classes [Sh:88] and for universal classes [Sh:300] led to the conclusion that an axiomatic approach provided the best setting for developing a theory of wider application. In the first chapter we describe the axioms on which the remainder of the article depends and give some examples and context to justify this level of generality. The study of universal classes takes as a primitive the notion of closing a subset under functions to obtain a model. We replace that concept by the notion of a prime model. We begin the detailed discussion of this idea in Chapter II. One of the important contributions of classification theory is the recognition that large models can often be analyzed by means of a family of small models indexed by a tree of height at most omega. More precisely, the analyzed model is prime over such a tree. Chapter III provides sufficient conditions for prime models over such trees to exist.