Interpreting groups and fields in simple, finitary AECs (Q435188)

The paper under review contributes to geometric stability theory in abstract elementary classes (AECs). The main result is a generalisation of a theorem in (elementary first-order) stability theory due to \textit{E. Hrushovski} [Ann. Pure Appl. Logic 45, No. 2, 139--155 (1989; Zbl 0697.03023)], namely the geometric classification of almost orthogonal (non-orthogonal) regular types, as either coming from a regular action of an abelian group, the action of the affine linear group on \(\mathbb{A}^1(K)\) or else the action of \(\mathrm{PGL}_2(K)\) on \(\mathbb{P}^1(K)\), for \(K\) an algebraically closed field. Previously, Hrushovski's result was generalised to non-elementary contexts in work of \textit{T. Hyttinen} et al. [J. Math. Log. 5, No. 1, 1--47 (2005; Zbl 1082.03033)]. The present paper differs from this work in that it assumes much less homogeneity for the AEC \((\mathbb{K},\preccurlyeq_{\mathbb{K}})\) under consideration, pushing the context even further away from a framework where compactness holds. This gain in generality is levelled against strong hypotheses on the independence notion (associated to Lascar non-splitting) in \((\mathbb{K},\preccurlyeq_{\mathbb{K}})\). The authors work in a simple, superstable finitary AEC \((\mathbb{K},\preccurlyeq_{\mathbb{K}})\), with monster model \(\mathfrak{M}\). They first gather some facts about independence and regular (Lascar strong) types in this context. Given an unbounded Lascar strong type \(p\) over a small set \(A\) with set of realisations \textbf{P}, denote by \textbf{P}\(/E\) the geometry associated with the natural pregeometry on \textbf{P}. Consider the following setting: \textbf{Q} is an \(A\)-invariant subset of \(\mathfrak{M}\), and there is \(0<n<\omega\) such that \noindent (i) for any independent sequence \((a_1,\ldots,a_n)\) in \textbf{P} and any finite \(C\subseteq \mathbf Q\), one has \(\dim(a_1,\ldots,a_n/A)=\dim(a_1,\ldots,a_n/A\cup C)\), and \noindent (ii) there is an independent sequence \((a_1,\ldots,a_{n+1})\) in \textbf{P} and some finite subset \(C\) of \textbf{Q} such that one has the inequality \(\dim(a_1,\ldots,a_{n+1}/A)>\dim(a_1,\ldots,a_{n+1}/A\cup C)\). Let \textbf{G} be the group of `elementary permutations' of \textbf{P}\(/E\) over \textbf{Q}. The action of \textbf{G} is shown to be transitive on independent \(n\)-tuples (\(n\) being maximal with this property), and an element of \textbf{G} is determined by its restriction to any \(n+1\)-dimensional subset of \textbf{P}. It follows by results of Hyttinen et al. [loc. cit.] that \textbf{G} is interpretable in \(\mathfrak{M}\) (in a suitable sense) over a finite set of parameters. Due to the absence of a good theory of \(\mathfrak{M}^{\mathrm{eq}}\) in this framework, in order to prove their classification result, the authors need to strengthen (ii) to: \noindent (ii\('\)) There is an independent sequence \((a_1,\ldots,a_{n+1})\) in \textbf{P} and some finite subset \(C\) of \textbf{Q} such that \((a_1,\ldots,a_n)\) dominates \((a_1,\ldots,a_{n+1})\) over \(A\cup C\). Under these assumptions, the main result is proved: \(\mathfrak{M}\) interprets a group \(G\) which acts on \textbf{P}\(/E\). Furthermore, either \(\mathfrak{M}\) interprets a non-classical group (i.e., a non-abelian group carrying an \(\omega\)-homogeneous pregeometry -- the existence of such a group is an open problem), or \(n\in\{1,2,3\}\) and, if \(n=1\), \(G\) is abelian and acts regularly on \textbf{P}\(/E\), if \(n=2\), the action of \(G\) on \textbf{P}\(/E\) is isomorphic to the action of the affine linear group acting on \(\mathbb{A}^1(K)\), and if \(n=3\) it is isomorphic to the action of \(\mathrm{PGL}_2(K)\) on \(\mathbb{P}^1(K)\), where \(K\) an algebraically closed field. The paper ends with a discussion of examples, one being Pappian projective planes, thus connecting their theorem to a very classical coordinatisation result.