scientific article; zbMATH DE number 3859231

From MaRDI portal
Publication:3328623





The two main results of this paper concern integer-valued entire functions mapping a subgroup \(\Gamma\) of \(\mathbb{C}^n\) to an algebraic number field \(K\). Let \(\ell \geq 1\) be the rank of \(\Gamma\) and let \(\gamma_1,\ldots,\gamma_{\ell}\) be generators for \(\Gamma\). Let \(f(z)\) be entire on \(\mathbb{C}^n\) and suppose there exist positive constants \(c,\kappa\) such that \(| f(z)| \leq \exp(c| z|^{\kappa})\) whenever \(| z| \geq 1\); suppose further that for any \(S\geq 1\) and any integers \(s_1,\ldots,s_{\ell}\) of absolute values at most \(S\) the value \(f(s_1\gamma_1+\cdots+s_{\ell}\gamma_{\ell})\) is an integer of \(K\) with all its conjugates of absolute values at most \(\exp(cS^{\kappa}).\) In the case \(n=1\) it is proved that these hypotheses force \(f(z)\) to be a polynomial provided \(\kappa =\ell\) and \(c\) is sufficiently small in terms of \(\gamma_1,\ldots,\gamma_{\ell}\) and \(K\). It is also remarked that this result becomes false if \(c\) is not small; in particular the exponent \(\kappa =\ell\) is best possible. In fact in certain cases with \(\ell =1,2\) even the correct constant \(c\) is known; see the first author's paper [(*): Invent. Math. 63, 495--506 (1981; Zbl 0461.10028)] for recent results and further references. For \(n>1\) some extra conditions on \(\Gamma\) are needed; assume that \(\ell =2n\) and that the generators \(\gamma_1,\ldots,\gamma_{2n}\) of \(\Gamma\) have algebraic coordinates and are linearly independent over \(\mathbb{R}\). It is then proved that the above hypotheses force \(f(z)\) to be a polynomial provided \(\kappa =2\) and \(c\) is sufficiently small in terms of \(\gamma_1,\ldots,\gamma_{2n}\) and \(K\). This result is similarly best possible. The proofs involve constructing auxiliary functions along the lines introduced by Gel'fond in his book and developed by \textit{M. Waldschmidt} [Acta Math. Acad. Sci. Hung. 31, 21--25 (1978; Zbl 0381.10029)], together with the functional equations technique of the first author (*). In the case \(n>1\) the latter technique meets with certain difficulties which the authors overcome by the ingenious expedient of restricting to a cone. [For the entire collection see Zbl 0504.00005.]











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 Q3328623)