scientific article; zbMATH DE number 3933176

From MaRDI portal
Publication:3708085





As usual \(S\) and \(T\) denote the sets of Pisot and Salem numbers, respectively. The authors give a number of results of the form: the real number \(\theta >1\) is in \(S\) \((S\cup T)\) if and only if there is a \(\lambda >0\) so that \(\| \lambda \theta^ n\|\), \(n\geq 0\) is sufficiently small. The prototype is \textit{Ch. Pisot}'s result [Ann. Sc. Norm. Super. Pisa, II. Ser. 7, 205--248 (1938; Zbl 0019.15502)] that \(\sum \| \lambda \theta^ n\|^ 2<\infty\) implies that \(\theta\) is in \(S\). They begin by giving an exposition of the methods and results of \textit{D. G. Cantor} [Acta Arith. 34, 43--55 (1977; Zbl 0373.10018)] and then give a refinement of his Theorem 2.1 by the same methods. In section 4, they note that there is an error in Cantor's proof of his Theorem 2.7 (the inequality (3.6) on p. 51 is incorrect). They modify the proof of this theorem to prove that \(\| \lambda \theta^ n\| =o(n^{-1/2})\) implies that \(\theta\) is in \(S\). This improvement of Pisot's result is, however, more easily derived from the remark of \textit{R. Salem} [Algebraic numbers and Fourier analysis. Boston: D. C. Heath and Co. (1963; Zbl 0126.07802), p. 12] that \(\sum^{n}_{k=1}k \| \lambda \theta^ k\|^ 2=o(n)\) implies that \(\theta\) is in \(S\). The boundary of the method seems to be Theorem 5.1 which states that if there is an \(a<1/(2(1+\theta)^ 2)\) and an \(n_ 0\) so that \(\| \lambda \theta^ n\| <a/n^{1/2}\) for \(n\geq n_ 0\) then \(\theta\) is in \(S\). Hardy's question (Pisot's conjecture), whether \(\| \lambda \theta^ n\| =o(1)\) suffices to show that \(\theta\) is in \(S\), is still open.











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 Q3708085)