Groups definable in local fields and pseudo-finite fields (Q1320034): Difference between revisions

This paper is concerned with groups which are first-order definable in some field \(F\). An example of an \(F\)-definable group is just provided by the group \(H(F)\) of \(F\)-rational points of an (abstract) algebraic group \(H\) over \(F\). It is well known that, if \(F\) is algebraically closed, then the converse is also true, in the sense that every \(F\)-definable group \(G\) is isomorphic in an \(F\)-definable way to \(H(F)\) for a suitable algebraic group \(H\) over \(F\); this follows from a theorem of A. Weil saying that any algebraic group over an algebraically closed field can be recovered by birational data [see \textit{E. Bouscaren}'s paper in Model theory of groups, Notre Dame Math. Lect. 11, 177-185 (1989; Zbl 0794.03050)]. Here the authors deal with \(F\)-definable groups when \(F\) is the real field \(\mathbb{R}\), or some \(p\)-adic field \(\mathbb{Q}_ p\), or a pseudofinite field (so an infinite model of the theory of finite fields). Some recent works of the second author prove that a group \(G\) definable in \(\mathbb{R}\) or in \(\mathbb{Q}_ p\) can be definably equipped with a structure of a real or \(p\)-adic Nash group [see J. Pure Appl. Algebra 53, 239-255 (1988; Zbl 0662.03025), and Arch. Math. Logic 29, 1-7 (1989; Zbl 0687.03016)]. Theorem A here shows that, if \(G\) is a Nash group over \(F = \mathbb{R}\) or \(\mathbb{Q}_ p\), then there exists an algebraic group \(H\) over \(F\) and a Nash isomorphism between neighbourhoods of the identity in \(G\) and \(H(F)\). It is observed that this local Nash isomorphism does not always lift to a global Nash isomorphism between \(G\) and \(H(F)\). But Theorem B ensures that, if \(G\) is a (topologically) connected Nash group over \(\mathbb{R}\), then, for some algebraic group \(H\) over \(\mathbb{R}\), there is a Nash isogeny between \(G\) and the connected component of \(H(\mathbb{R})\) (isogeny means a surjective homomorphism with a finite central kernel). Finally, the authors treat pseudofinite fields. Theorem C shows that, if \(G\) is a group definable in such a field \(F\), then there exists an algebraic group \(H\) over \(F\) and an \(F\)-definable isogeny between subgroups of finite index in \(G\) and in \(H(F)\). The main obstacle to overcome with respect to the algebraically closed case is that real, \(p\)-adic and pseudofinite fields are not stable (in the model theoretic sense), while algebraically closed fields are even strongly minimal. But the authors show that, if \(F\) is \(\mathbb{R}\), or \(\mathbb{Q}_ p\), or a pseudofinite field, then \(F\) -- as a substructure of its algebraic closure -- satisfies some assumptions ensuring that independence, dimension and generic points (hence some basic tools in the strongly minimal case) still make sense; a field with these properties is called a geometric field. On this basis, Theorems A and B are shown by using the topological field structure of \(\mathbb{R}\) and \(\mathbb{Q}_ p\), while the proof of Theorem C needs some more results of local stability theory. The paper is very well written and contains a clear introduction explaining the logical background to non-model theorists; a lexicon translating the model-theoretic language in the context of algebraic geometry is also provided.