The Manin-Mumford conjecture and the model theory of difference fields (Q5945398)

The Manin-Mumford conjecture, in the general form proved by Raynaud in 1983, states that if \(A\) is an Abelian variety over a number field \(K\), \(X\) is a subvariety (and \(T(A)\) denotes the group of torsion points of \(A\) over the algebraic closure of \(K\)), then there exists a finite number \(M\) of subvarieties \(c_i + A_i\) (\(i < M\)) of \(X\), all translates of group subvarieties \(A_i\) of \(A\), such that \(T(A) \cap X = \bigcup_i (c_i + T(A_i))\). This paper gives a new proof of this result (and of other related statements), extends its validity to arbitrary commutative algebraic groups \(A\) and, above all, provides effective bounds on \(M\), \[ M \leq c (dg X)^e, \] where \(c\) and \(e\) depend on \(A\) but not on \(X\); indeed, \(c\) and \(e\) are written down explicitly, and are doubly exponential in some natural parameters associated with \(A\). Although both number-theoretic and algebraic methods are involved in the proof, the spirit of this approach is essentially model-theoretic, in particular it largely owes to the model theory of (existentially closed) difference fields. Actually the connection with difference algebra was already present in the classical number-theoretic proofs of the Manin-Mumford conjecture. In fact, all of them dealt first with the points of order relatively prime to a given prime \(p\) and observed that, for some constant \(d = d(K, A)\) and for all prime-to-\(p\) torsion points \(t\) of (the Abelian variety) \(A\), \(t\) and \(dt\) are in the same conjugacy class over \(K\). Consequently, if \(t \in X\), then \(dt \in X\) as well, so that the resulting setting can be interpreted as intersecting \(X\) with solutions to a difference equation \(\sigma (x) = d x\) for a suitable automorphism \(\sigma\). Also in this new approach the first step of the proof (the only part using number theory in a relevant way) shows that the torsion points are contained in a group defined by an appropriate difference equation, arising from characteristic equations of Frobenius maps, lifted to characteristic 0. This leads to an analysis of the Abelian group definable in existentially closed difference fields (i.e. models of the model companion of the theory of fields with an automorphism) in characteristic 0. Here the paper extends and specializes the general model theory of difference equations in characteristic 0 developed in a previous paper written by the author together with Z. Chatzidakis. The structure of definable Abelian groups of finite dimension is investigated by using methods of geometric stability theory (sometimes generalized to finite \(S1\)-rank). Moreover, bounds on the number of solutions to a finite set of difference equations are provided. As already said, these tools are used to deduce a full model-theoretic proof of the Manin-Mumford conjecture and to provide the effective bounds mentioned before. Other notable applications are discussed. In particular, the author deals with the Tate-Voloch conjecture stating that the torsion points on an Abelian variety \(A\) over \({\mathbb C}_p\) that do not lie on a subvariety \(V\) of \(A\) are bounded away from that variety. A positive answer is obtained as an almost immediate consequence of the solution of the Manin-Mumford conjecture, under some suitable restrictions on \(A\). The author informs that a subsequent work of Thomas Scanlon removed these last constraints. More generally, the author describes the genesis of the whole paper: in fact, his first proof of the Manin-Mumford conjecture was found in 1994. Since then, some relevant progress in model theory has intervened (including, for instance, the development of the theory of simple unstable theories and the Scanlon work just recalled), and new hints and intriguing connections (like the reference to the Tate-Voloch conjecture) have enriched the original 1994 proof. The present version, as it can be read today, is still a fertile and enjoyable source of deep and illuminating ideas on modern model theory and its lively connections with number theory.