scientific article; zbMATH DE number 3904689

From MaRDI portal
Publication:3682563





This is an announcement of the author's profound generalization to several variables of the well-known transcendence criterion of Gelfond. For \(n\geq 1\) and \(0\leq k<n\) let \(a\), \(b\) be real numbers with \(0<b<a<b(k+2)/(k+1)\). Let \(\tau(N)\), \(\delta(N)\) be increasing functions with \(\tau(N)\ge \delta(N)\ge 1\) and \(\delta(N)\to \infty\) as \(N\to \infty\). Let \(\vartheta =(\vartheta_1,\ldots,\vartheta_n)\) be a point of \(\mathbb{C}^n\) with the following property. For each sufficiently large integer \(N\) there exist polynomials \(P_{1,N},\ldots,P_{m(N),N}\) in \(\mathbb{Z}[X_1,\ldots,X_n]\), of degrees at most \(\delta(N)\) and tailles at most \(\tau(N)\), having only finitely many common zeros within \(\exp(-\tau(N)\delta(N)^{k+a})\) of \(\vartheta\), whose absolute values at \(\vartheta\) are not all zero and at most \(\exp(-\tau(N) \delta(N)^{k+b})\). Then at least \(k+1\) of \(\vartheta_1,\ldots,\vartheta_n\) are algebraically independent over \(\mathbb{Q}\). The author gives some applications as follows. For \(\ell \geq 1\), \(m\geq 1\) let \(x_1,\ldots,x_{\ell}\), \(y_1,\ldots,y_m\) be complex numbers such that for each \(\varepsilon >0\) we have \[ | \sum^{\ell}_{i=1}\lambda_ i x_i| \ge \exp(-X^{\varepsilon}),\quad | \sum^{m}_{j=1}\mu_j y_j| \ge \exp (-X^{\varepsilon}) \] for all integers \(\lambda_1,\ldots,\lambda_{\ell}\), \(\mu_1,\ldots,\mu_m\) with \(X=\max (| \lambda_1|,\ldots,| \lambda_{\ell}|)=\max (| \mu_1|,\ldots,| \mu_m|)\) sufficiently large. Let \(K\) be the field obtained by adjoining all the \(\exp(x_i y_j)\) to \(\mathbb{Q}\), let \(K'=K(x_1,\ldots,x_{\ell})\), and let \(K=K'(y_1,\ldots,y_m)\). Then the transcendence degrees of \(K\), \(K'\), \(K\) over \(\mathbb{Q}\) are at least \((\ell m-\ell -m)/(\ell +m)\), \((\ell m-m)/(\ell +m)\), \(\ell m/(\ell +m)\), respectively. A striking consequence is that if \(\alpha\ne 0,1\) and \(\beta\) are algebraic numbers with \(\beta\) of degree \(d\ge 2\), then at least \([d/2]\) of the numbers \(\alpha^{\beta},\ldots,\alpha^{\beta^{d-1}}\) are algebraically independent over \(\mathbb{Q}\). These are all very substantial improvements on earlier results; in particular every previous lower bound for the last mentioned transcendence degree increased only logarithmically with \(d\). The above algebraic independence criterion is proved, in slightly stronger form, in the author's important paper Critères d'indépendance algébrique shortly to appear in [Publ. Math., Inst. Hautes Étud. Sci. 64, 5--52 (1986; Zbl 0615.10044)]. This also contains proofs of the applications as well as full historical references. Finally the author states a quantitative version of his criterion, which yields an appropriately defined algebraic independence measure for \(\vartheta_1,\ldots,\vartheta_n\); but this is too technical to quote here. More details appeared in [Théorie des nombres, Sémin. Delange-Pisot-Poitou, Prog. Math. 59, 219--233 (1985; Zbl 0567.10034)], see the following review.




Cited in
(29)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3682563)