Algebraic curves and multiplicative equations. (Q930542)

The main object of the paper under review is an algebraic curve \(C\) lying in a multiplicative torus \(A=\mathbb G_{\text{{m}}}^ g\) defined over \(\bar\mathbb Q\). The author's goal is to study the intersection of \(C(\bar\mathbb Q)\) with the union of \(H(\bar\mathbb Q)\) where \(H\) runs over the set of all algebraic subgroups of \(A\) of codimension 2. More precisely, he is interested in describing \(C\) such that this intersection is finite. Towards this end, \(C\) is called \textit{transversal} if it is not contained in any translate of a proper subtorus of \(A\), and \textit{weakly transversal} if it is not contained in any proper algebraic subgroup of \(A\). The main result of the paper (Theorem 1.2) states that if \(C\) is weakly transversal then the above mentioned intersection is finite. This generalizes an earlier result by \textit{E.~Bombieri, D.~W.~Masser} and \textit{U.~Zannier} [Int. Math. Res. Not. 1999, No. 20, 1119--1140 (1999; Zbl 0938.11031)] who proved the same statement for transversal curves, stated the assertion of Theorem 1.2 as a conjecture, and proved it for \(g\leq 5\) in a later publication [Trans. Am. Math. Soc. 358, 2247--2257 (2006; Zbl 1161.11025)]. Furthermore, using more recent results of the same authors, Theorem 1.2 turned out to be extendable to the case where \(\bar\mathbb Q\) is replaced with any algebraically closed field of characteristic zero. These results can be reformulated so that they could be viewed as toric analogues of various specialization theorems (Demjanenko, Manin, Silverman). They are also special cases of more general conjectures proposed by Zilber (where \(A\) is replaced with any semi-abelian variety) and Pink (who formulates the problem in terms of mixed Shimura varieties). The proofs are motivated by the work of \textit{G.~Rémond} and \textit{E.~Viada} [Int. Math. Res. Notices 35, 1915--1931 (2003; Zbl 1072.11038 )] on a particular case of the Zilber conjecture (\(A=E^ g\), \(E\) is an elliptic curve). They are relied on a variant of the Vojta inequality.