Über eine Vermutung von S. Chowla und H. Walum (Q761493)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Über eine Vermutung von S. Chowla und H. Walum
scientific article

    Statements

    Über eine Vermutung von S. Chowla und H. Walum (English)
    0 references
    0 references
    0 references
    1985
    0 references
    Let \(\psi_k(\cdot)\) denote the 1-periodic Bernoulli polynomial of order \(k\in\mathbb{N}\). In generalization of Dirichlet's classical divisor problem, \textit{S. Chowla} and \textit{H. Walum} [Norske Vid. Selsk. Forh. 36, 127--134 (1963; Zbl 0135.10703)] conjectured that for \(x\to \infty\) the estimate \[ \sum_{n\leq x^{1/2}}n^{\alpha}\psi_ k(x/n) = O(x^{\vartheta (\alpha)});\ \vartheta (\alpha)=1/4+\alpha /2 + \varepsilon \tag{*} \] still holds for real \(\alpha\geq 0\) and arbitrary \(\varepsilon >0\). \textit{S. Kanemitsu} and \textit{R. Sitaramachandrarao} [J. Number Theory 20, 255--261 (1985; Zbl 0467.10031)] established the conjecture in case of \(k\geq 2\) and \(\alpha\geq 1/2\). \textit{W. G. Nowak} [Bull. Number Theory Rel. Top. 7, No. 1, 1--10 (1982; Zbl 0491.10013)] improved their results in the remaining case \(k\geq 2\) and \(0\leq \alpha <1/2\), e.g. \(\vartheta (0)=2/7=0,2857\ldots\). The objective of the paper is to present a uniform approach for estimating sums like (*) in the case \(k\geq 2\). The corresponding exponential sum is estimated with the aid of van der Corput's theory of exponent pairs in the simplified form by \textit{E. Phillips} [Q. J. Math., Oxf. Ser. 4, 209--225 (1933; Zbl 0007.29801)] using an argument of \textit{H.-E. Richert} [Math. Z. 56, 21--32 (1952; Zbl 0046.25002)]. As an application of the theorem one gets slightly refinements over the results of W. G. Nowak (loc. cit.), e.g. \(\vartheta (0)=109/382=0,2853\ldots\). For \(k\geq 2\), \(\alpha\geq 1/2\) the conjecture is proved within a few lines. \{Note: W. G. Nowak pointed out to the author that the special case \(k=2\), \(\alpha =0\) in (*) is implicitly Wigert's divisor problem. The best known result about the corresponding exponent hitherto is that of \textit{H. Y. Peng} [Sci. Record 1, 69--72 (1942; Zbl 0060.10406)] namely \(2/7\) and can be replaced by \(109/382\).\}
    0 references
    0 references
    0 references
    0 references
    0 references
    Chowla-Walum conjecture
    0 references
    generalized divisor problem
    0 references
    Bernoulli polynomial
    0 references
    Dirichlet's classical divisor problem
    0 references
    exponent pairs
    0 references
    Wigert's divisor problem
    0 references