Around the Chevalley-Weil theorem (Q2154809)

The Chevalley-Weil theorem is a powerful tool in Diophantine Geometry. It roughly asserts that given an unramified morphism of finite degree \(\pi \colon W \to V\) of projective algebraic varieties defined over a number field \(k\), one can lift the rational points \(p \in V(k)\) to points of \(W\) defined over a fixed number field. One key point of this result is the assumption that the morphism is ``unramified'', since otherwise, if \(\operatorname{deg}(\pi) >1\) and if \(V\) has enough rational points, then one would expect that the field of definitions of the fibers \(\pi^{-1}(p)\) vary with \(p\), so as to generate a field of infinite degree. A proof of this result was first given by \textit{C. Chevalley} and \textit{A. Weil} [C. R. Acad. Sci., Paris 195, 570--572 (1932; Zbl 0005.21611; JFM 58.0182.04)] in the case of curves. Since then, many proofs have appeared in the literature, also for arbitrary dimension (e.g. in the works of Bombieri and Gubler, Lang, Serre, etc.). The main goal of the present note is to illustrate a self-contained proof of this theorem with a rather different presentation and assumptions of purely topological content (see Theorem 1.1). In order to this, the authors discuss and compare various concepts of ``ramification''. They finally adopt a purely topological notion of ramification via the notion of ``topological cover'', with which they are able to exploit the absence of it and deducing it for the number fields after specialization. In previous proofs, the notion ``unramified'' was always formulated in an algebraic way (e.g.~ ``étale'' or ``\(\Omega^1_{W/V} = 0\)''). Finally, the authors apply Theorem 1.1 to the study of solutions of generalized Fermat equations. Even though this note does not contribute to substantially new results, the notion of topological cover is new in this context, and it provides a more accessible statement of the Chevalley-Weil theorem which is easier to use for applications.