Fractional parts of powers of rationals. (English) Zbl 0298.10018

Let \(a\) and \(q\) be any relatively prime integers with \(a >q\ge 2\) Mahler proved in 1957 that for any \(\varepsilon >0\) there exist only finitely many positive integers \(n\) such that \(\vert(a/q)^n\Vert < e^{-\varepsilon n}\); here \(\Vert x\Vert\) denotes the distance of \(x\) from the nearest integer [see K. Mahler, Mathematika 4, 122–124 (1957; Zbl 0208.31002)]. The authors establish an effective theorem which is slightly weaker: There exist effectively computable numbers \(N\) and \(\eta\), with \(0< \eta <1\), such that \(\vert(a/q)\Vert^n > q^{-\eta n}\) for all integers \(n\ge N\). It is not possible from this result to derive that the number \(g(k)\) in Waring’s problem is given by \[ 2^k + \left[\left(\frac32\right)^k\right] - 2 \] for sufficiently large \(k\), as follows from Mahler’s result. The theorem in the present paper is based on the following \(p\)-adic analogue of Baker’s result in [Acta Arith. 24, 33–36 (1973; Zbl 0261.10025)]. Let \(p\) be a prime and \(a\) any non-zero integer. There is an effectively computable number \(c=c(a,p)\) such that, for any \(\delta\) with \(0< \delta \le \tfrac12\), the inequalities \[ 0 < \vert a^n - b\vert_p < \delta^{c\log \vert 2b\vert} e^{-\delta n} \] have no solution in integers \(b\) and \(n>0\). In the proof a new device has been used to overcome the fact that the \(p\)-adic exponential function converges only in a certain neighbourhood of the origin.


11J54 Small fractional parts of polynomials and generalizations
11J61 Approximation in non-Archimedean valuations
Full Text: DOI


[1] Mahler, Mathematika 4 pp 122– (1957)
[2] Coates, Acta Arith. 15 pp 279– (1969)
[3] Baker, Mathematika 13 pp 204– (1966)
[4] Baker, Acta Arith. 21 pp 117– (1972)
[5] Baker, Acta Arith. 24 pp 33– (1973)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.