Categorical Horn theories and modules (Q662650)

As a part of a research program dating back 35 years, the author considers the relationships between arbitrary categorical Horn theories and theories of modules. A first-order theory is \textit{Horn} if the class of models is closed under reduced products, equivalently if it is axiomatized by Horn sentences. A theory is \textit{categorical} if it has a unique model up to isomorphism in some/all cardinalities greater than the cardinality of the language. The author has already classified the models of the categorical universal Horn theories (quasivarieties) [Algebra Logic 14(1975), 86--111 (1976); translation from Algebra Logika 14, 145--185 (1975; Zbl 0319.08004)] and the categorical positive Horn theories [Algebra Logic 19, 443--455 (1981); translation from Algebra Logika 19, 683--700 (1980; Zbl 0491.03010)], and has developed a reasonable general structure theory for the models of arbitrary categorical Horn theories [Algebra Logic 19, 377--400 (1981); translation from Algebra Logika 19, 582--614 (1980; Zbl 0491.03011); Algebra Logic 49, No. 6, 526--538 (2011); translation from Algebra Logika 49, No. 6, 782--802 (2010; Zbl 1220.03013)]. Key concepts for the study of Horn theories are positive primitive formulas (existential quantifications of conjunctions of atomic formulas) and primitive normal theories, that is, theories such that the sets defined by two instances \( \varphi(\overline{x},\overline{b}) \) and \( \varphi(\overline{x},\overline{c}) \) of a positive primitive formula are equal or disjoint. One key result of the most recent of these papers is that if a categorical Horn theory does not interpret any non-trivial group by a positive primitive formula, then it is (essentially) a universal Horn theory. This paper considers the case where the theory interprets an abelian group by a positive primitive formula. In \S3, the author describes the structure of models of categorical Horn theories of modules. In \S4, he considers expansions of abelian groups which are primitive normal. In \S5 he deals with the question ``When is a categorical Horn theory primitively equivalent up to constants to a theory of a module?'' Such a theory must at the very least positively interpret an affine structure, so the results of the preceding two sections apply. Palyutin finds some sufficient conditions, by restricting the language or the length of chains of pp-definable sets, that force a categorical affine Horn theory to be the theory of a module. There are no necessary conditions, and the larger question remains open. This paper and its predecessors form an interesting body of work which would benefit a great deal by some modernization. There is a very rich literature on Horn theories in general. A good starting point would be Chapter 9 of \textit{W.\ Hodges}' text [Model theory. Cambridge: Cambridge University Press (1993; Zbl 0789.03031)] and the references therein, although Hodges says almost nothing about the categorical case. Palyutin's only reference for the model theory of modules is the seminal paper [\textit{M. Ziegler}, Ann. Pure Appl. Logic 26, 149--213 (1984; Zbl 0593.16019)]. Most of the results of \S3 can be found in, or are simple consequences of, basic facts that can be found in \textit{M. Prest}'s book [Model theory and modules. Cambridge etc.: Cambridge University Press (1988; Zbl 0634.03025)] The work in \S4 and \S5 should be recast in terms of ``abelian structures in the sense of Fisher'' [\textit{E. R. Fisher}, Lect. Notes Math. 616, 270--322 (1977; Zbl 0414.18001)], which receives good expositions in [Prest, loc. cit.] and [\textit{M. Prest}, Purity, spectra and localisation. Cambridge: Cambridge University Press (2009; Zbl 1205.16002)]. The proof of the key Proposition 3 of \S4 can be easily adapted to show that, if \(A\) is a primitive normal structure on an underlying abelian group, then \(A\) is, up to constants, a 1-sorted abelian structure. The ring that Palyutin introduces after Proposition 3 of \S4 is the ring of definable scalars, which has been studied extensively since it was first introduced in the very early days of the model theory of modules. There is extensive coverage, with references, in [Prest (2009), loc. cit.]. In the context of abelian structures the type of construction that Palyutin uses in Example 1/Proposition 4 of this section is well-known: an abelian structure may fail to be equivalent to a module simply by not having enough operations defined on it to capture all the definable relations. The main question of \S5 reduces to: ``Is every one-sorted categorical abelian structure positively equivalent to the natural module over its ring of definable scalars?'' More general approaches to the interpretations of theories of modules or abelian structures in each other are functorial, and [\textit{M. Prest}, Ann. Pure Appl. Logic 88, No. 2--3, 193--215 (1997; Zbl 0946.03044)] gives excellent coverage of this material and an extensive list of references.