Minimal first-order structures (Q639691): Difference between revisions

The paper answers positively a conjecture raised by Anand Pillay in his Ph.D. thesis about the number \(I(\aleph_0, T)\) of countable pairwise non-isomorphic models of the elementary diagram \(T\) of an infinite countable structure \(C\). Pillay proved that \(I(\aleph_0, T) \geq 4\) but conjectured that \(I(\aleph_0, T)\) is actually infinite. The author points out that the crucial case to be examined towards a positive solution of the conjecture is that of regarding a minimal \(C\) (where `minimal' means that every definable subset of \(C\) is either finite or cofinite). Thus he turns his attention to these structures and singles out two key behaviors in their class. In order to introduce them, let us denote by \(U\) a big saturated elementary extension of \(C\) and by \(p\) the unique non-algebraic 1-type over \(C\). Then the dichotomy theorem says that either {\parindent=6mm \begin{itemize}\item[(i)] \(p(U)\) is a pregeometry with respect to the semi-isolation operator, or \item[(ii)] there is some infinite subset \(C_0\) of \(C\) directing a type over a finite \(E \subseteq U\). \end{itemize}} Recall that the semi-isolation operator is the operator sending every \(A \subseteq p(U)\) into the set of all \(a \in p(U)\) which are semi-isolated over \(A\) (meaning that some formula \(\varphi(x) \in \text{tp}(a/A \cup C)\) implies the whole \(p\)). Moreover, an infinite \(C_0 \subseteq C\) is said to direct a type over \(E\) basically if there is an \(E\)-definable quasi-order \(\preceq\) on \(U\) such that \(C_0\) is downward closed with respect to \(\preceq\) and, whenever \(\text{tp}(b/aE)\) is a non-algebraic 1-type finitely satisfiable in \(C_0\), then \(b \preceq a\). Strongly minimal structures satisfy (i), while both \((\omega, \leq)\) and \((\omega + \omega^\star, \leq)\) exemplify (ii). Pillay's conjectures is a consequence of this dichotomy theorem. In fact, it is shown that (i) implies \(I(\aleph_0, T) \geq \aleph_0\), while, by a parallel result of the author, \(I(\aleph_0, T) = 2^{\aleph_0}\) when (ii) holds.