Rarified sums of the Thue-Morse sequence (Q4699608)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1363755
Language Label Description Also known as
default for all languages
No label defined
    English
    Rarified sums of the Thue-Morse sequence
    scientific article; zbMATH DE number 1363755

      Statements

      Rarified sums of the Thue-Morse sequence (English)
      0 references
      0 references
      0 references
      17 November 1999
      0 references
      summatory functions
      0 references
      Thue-Morse sequence
      0 references
      Let \((t_n)_{n\geq 0}\) be the Thue-Morse sequence with values \(\pm 1\) (i.e., \(t_n= (-1)^{s(n)}\), where \(s(n)\) is the sum of the binary digits of \(n\)). Define \(S_{q,i}(n)= \sum_{\substack{ 0\leq j< n\\ j\equiv i(q)}} t_j\). \textit{D. J. Newman} proved in 1969 [Proc. Am. Math. Soc. 21, 719-721 (1969; Zbl 0194.35004)] that \(S_{3,0}(n)> 0\), which indicates that \(t_{3n}\) is ``more often'' equal to \(+1\) (of course the density of the set \(\{n\); \(t_{3n}=+1\}\) is equal to \(1/2\)). Actually Newman proved that \[ 0< \alpha n^{\frac{\log 3}{\log 4}}< S_{3,0}(n)< \beta n^{\frac{\log 3}{\log 4}}. \] This result was made more precise by \textit{J. Coquet} [Invent. Math. 73, 107-115 (1983; Zbl 0528.10006)] who proved that \[ S_{3,0}(n)= n^{\frac{\log 3}{\log 4}} \psi\biggl( \frac{\log n}{\log 4} \biggr)+ O(1), \] where \(\psi\) is a continuous nowhere differentiable function with period 1. Asymptotic expressions of \(S_{q,i}(n)\) were studied by many authors, but the difficult problem is whether \(S_{q,i}(n)> 0\). The author of the paper under review prove the following nice results: -- if \(3|q\) or if \(q= 4^N+1\), then \(S_{q,0}(n)> 0\) for \(n\) large enough (they indicate this has been extended to \(q= \frac{1}{3} (2^{2n+1}+1)\) by Leinfellner); -- the total numbers of primes \(p\) such that \(S_{p,0}(n)> 0\) for \(n\) large enough is \(o(x/\log x)\) as \(x\to +\infty\). Note that an interesting connection with the conjecture that there are only finitely many Fermat primes is given.
      0 references

      Identifiers