An estimate for the multiplicity of binary recurrences
Consider complex sequences \(\{ u_n\}_{n\in{\mathbb Z}}\) satisfying a binary recurrence relation \(u_n=c_1u_{n-1}+c_2u_{n-2}\) for certain complex numbers \(c_1,c_2\). Suppose that the companion polynomial \(X^2-c_1X-c_2\) has two roots \(\alpha\), \(\beta\) for which none of the numbers \(\alpha\), \(\beta\), \(\alpha/\beta\) is a root of unity. For a given number \(a\in{\mathbb C}\), denote by \(N(a)\) the number of \(n\in{\mathbb Z}\) with \(u_n=a\). \textit{H.-P. Schlickewei} was the first to obtain a uniform upper bound for \(N(a)\). In fact, he showed \(N(a)\leq 2^{2^{23}}\) for every \(a\) [Invent. Math. 129, 11-36 (1997; Zbl 0883.11008)]. His main tool was his own quantitative version of the two-dimensional so-called parametric subspace theorem (or rather parametric Roth's theorem). In the present paper, the author deduces a sharpening of Schlickewei's quantitative parametric Roth's theorem by using the sharper version of Roth's lemma that follows from a quantitative version of Faltings' product theorem. As a consequence, he obtains \(N(a)\leq 2^{57}\) for all \(a\in {\mathbb C}\). We mention that \textit{F. Beukers} and \textit{H.-P. Schlickewei} [Acta Arith. 78, 189-199 (1996; Zbl 0880.11034)] had already obtained the better bound \(N(a)\leq 61\) by another method based on hypergeometric functions rather than Roth's theorem.
This page was built for publication: An estimate for the multiplicity of binary recurrences
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4385788)
