Diophantine approximation on algebraic varieties (Q1975054)

The paper under review is a survey on recent achievements on arithmetic algebraic geometry due especially to Vojta and Faltings, as well as their links to classical Diophantine approximation results of Roth and Schmidt. The first chapter deals with Roth's Theorem. In 1954 Roth proved the following statement concerning rational approximation to algebraic numbers: let \(\alpha\) be an irrational real algebraic number, \(\varepsilon\) a positive real number. Then for all but finitely many pairs of integers \((p,q)\), \(|\alpha-p/q|>q^{-2-\varepsilon}\). The author gives the idea of the proof, stressing the main technical difficulty, i.e. a ``nonvanishing lemma'' for polynomials in several variables. Here two main approaches are possible: the original one (Roth's lemma) and the more modern ``Dyson's lemma'' in several variables, proved by \textit{H. Esnault} and \textit{E. Viehweg} [Invent. Math. 78, 445-490 (1984; Zbl 0545.10021)], revisited by \textit{M. Nakamaye} himself [Invent. Math. 121, 355-377 (1995; Zbl 0855.11036)]. The author gives a clear account of the latter, explaining its relation with Faltings' product theorem. The second part deals with Mordell conjecture (now Faltings' theorem) asserting the finiteness of rational points on an algebraic curve of genus greater than one. The link with Roth's theorem is provided by \textit{P. Vojta}'s proof of Faltings' theorem [Ann. Math. (2) 133, 509-548 (1991; Zbl 0774.14019)], which follows the main steps of classical proofs in Diophantine approximation. The author emphasises this link through the presentation of Vojta's generalization of Dyson's lemma to products of curves of arbitrary genus. The author also gives an overview of the subsequent proof by Faltings of the Lang conjecture on rational points on algebraic subvarieties of abelian varieties. Finally, a chapter is devoted to the new proof of the Subspace Theorem by Faltings and Wüstholz, leading to interesting generalizations. The Subspace Theorem, which is the natural generalization of Roth's theorem to higher dimension, provides a lower bound for the rational approximation to a hyperplane (or a family of hyperplanes) defined over the algebraic numbers. The new ideas of the Faltings-Wüstholz proof are clearly presented; they led to new results concerning approximations by rational points on an algebraic subvarieties as well as approximation to nonlinear subspaces. Since any clear account of these interesting new applications is lacking both in the Faltings-Wüstholz paper and in the article under review, the interested reader is referred to articles by \textit{J.-H. Evertse} and \textit{R. G. Ferretti}, especially [Int. Math. Res. Notes 25, 1295-1330 (2002)] and [\textit{R. G. Ferretti}, Compos. Math. 121, 247-262 (2000; Zbl 0989.11034)].