Nonessential extensions of complete theories (Q2266007)

An inessential extension T' of a complete theory T is a theory of the form \(Th(<M;\bar c>)\), where \(M\vDash T\) and \(\bar c\in M\). \textit{R. E. Woodrow} [J. Symb. Logic 43, 442-455 (1978; Zbl 0434.03022)] and \textit{M. G. Peretyat'kin} [Algebra Logika 19, 224-235 (1980; Zbl 0481.03015)] found pairs T and T' as above with \(n(T)<n(T')=\omega\), where n(T) denotes the number of countable models of T. The latter asked whether there are theories such that (1) \(n(T')<n(T)<\omega\), or (2) \(n(T')<n(T)=\omega\), and A. D. Tajmanov asked for those satisfying (3) \(n(T')<\omega\) \& \(n(T)=2^{\omega}\), or (4) \(n(T')=\omega\) \& \(n(T)=2^{\omega}.\) In the paper under review examples of the first three types are constructed. To have an idea of the constructions consider \(DLO^-\) (resp. DLO), the theory of dense linear orderings (without endpoints, resp.), and let T be the theory of a model of DLO, expanded by new constants \(a_ 0\), \(b_ 0\), \(a_ 1\), \(b_ 1\),... which form the following cut \(q: a_ 0<a_ 1<a_ 2<...<b_ 2<b_ 1<b_ 0\). Then in any model M of T the intervals \((a_ i,a_{i+1})\) and \((b_{i+1},b_ i)\) form models of DLO. So any countable such M is completely determined by its ''kernel'' q(M), the set of elements within the cut. Clearly, these kernels can be either empty, a singleton, or one of the four countable models of \(DLO^-\), and each of these occurs. Thus \(n(T)=6\). Now let \(c\in N\vDash T\) realize q. Consider \(T'=Th(<N;c>)\). Every model of T' is of the form \(<M;c>\) for some \(M\vDash T\) and \(c\in q(M)\). So, for q(M) itself there remain five isomoprhism types only (the empty one cannot occur). It can be easily checked that on those we have altogether nine ways to choose the constant c. Hence \(n(T')=9.\) Although this construction is not of the desired type, the one given in the paper under review is in fact a refinement of it: By putting a unary function and a predicate onto DLO in a clever way, the author ensures \(n(T)=6\) again, however in such a way that in every non-principal inessential extension T' of T (i.e. by constants realizing a non- principal type) there are altogether only five ways to pick the constants, whence \(n(T')=5\); this providing an example of type (1) (introducing n more predicates he gives also one with \(n(T)=n+5\) and \(n(T')=5)\). Appropriately applying this idea several times simultaneously, he arrives at examples of sort (2) and (3) as well. Unfortunately, the paper is neglectfully written, with a lot of misleading omissions, misprints and formal errors. Also the author does not even indicate, though claimed so in the introduction, how to obtain an example of type (4).