Bohr and Rogosinski abscissas for ordinary Dirichlet series (Q2378569)

From MaRDI portal
Revision as of 01:46, 23 February 2024 by RedirectionBot (talk | contribs) (‎Removed claim: author (P16): Item:Q1269148)
scientific article
Language Label Description Also known as
English
Bohr and Rogosinski abscissas for ordinary Dirichlet series
scientific article

    Statements

    Bohr and Rogosinski abscissas for ordinary Dirichlet series (English)
    0 references
    0 references
    0 references
    0 references
    13 January 2009
    0 references
    Let \(f(s)=\sum_{n=1}^\infty a_n n^{-s}\) be a Dirichlet series converging in the right half-plane \(\Pi= \{s\in \mathbb C \mid\Re s>0\}\) and \(f\in H^\infty(\Pi)\). Denote \(\|f\|=\sup_\Pi|f|\). By definition, the \textit{isometric Bohr abscissa} \textbf{b} is the infimum of \(\sigma\geq 0\) such that for all such \(f\), we have \(\sum_{n=1}^\infty |a_n| n^{-\sigma}\leq\|f\|\). For any plain domain \(G\) with convex hull \(\widetilde{G}\neq \mathbb C\), we define the \textit{isometric Bohr abscissa} \textbf{b}\((G)\) as the infimum of \(\sigma\geq 0\) such that for all \(f=\sum_{n=1}^\infty a_n n^{-s}\in H^\infty(\Pi)\) with \(f(\Pi)\subset G\), we have \(\sum_{n=2}^\infty |a_n| n^{-\sigma}\leq \text{dist}(a_1,\partial \widetilde{G})\). At last the \textit{Rogosinski abscissa} \textbf{r}\((G)\) is the infimum of \(\sigma'\geq 0\) such that, for all \(f=\sum_{n=1}^\infty a_n n^{-s}\) \(\in H^\infty(\Pi)\) with \(f(\Pi)\subset G\), we have \(P_k(\Pi_{\sigma'})\subset G\). Here \(P_k(s)=\sum_{n=1}^k a_n n^{-s}\) and \(\Pi_{\sigma'}= \{s\in C \mid\Re s>\sigma'\}\). If we only consider absolutely convergent Dirichlet series, then we obtain characteristics \textbf{b}\({}_a(G)\) and \textbf{r}\({}_a(G)\) instead of \textbf{b}\((G)\) and \textbf{r}\((G)\). Let \(U_1\) be the unit disc. For a plane domain \(G\), we call a point \(p\in \partial G\) \textit{a point of convexity} if \(p\in\partial \widetilde{G}\), and \textit{a regular point of convexity} if, in addition, there exists a disk \(U\) in \(G\) such that \(p\in \partial U\). Now we formulate the main results. In Section~2 the independence of the Bohr and Rogosinski constants of the domain \(G\) is investigated. It is proved that for bounded domains \(G\), we have \textbf{b}\((G)\leq{\mathbf b}(U_1)\), and equality is valid if \(G\) contains at least one regular point of convexity (Theorem~2.1). If \(\widetilde{G}\neq \mathbb C\), then \textbf{b}\({}_a(G)\leq {\mathbf b}(U_1)\) with equality if \(G\) contains at least one regular point of convexity (Theorem~2.2). For bounded convex domains, \textbf{r}\((G)={\mathbf r}(U_1)\) (Theorem~2.3), and for convex ones, \textbf{r}\({}_a(G)={\mathbf r}(U_1)\) (Theorem~2.4). At last, for bounded convex domains, \textbf{r}\((G)\leq {\mathbf b}(G)\), and for convex ones, \textbf{r}\({}_a(G)\leq {\mathbf b}{}_a(G)\). In Section~3 some estimates for the Bohr and Rogosinski constants are proved. It is shown that \textbf{b}\((G)\leq 1.7267\) for convex domains; for domains \(G\) with \(\widetilde{G}\neq\mathbb C\) we have \textbf{b}\({}_a(G)\leq 1.7267\) (Theorem 3.1). For \(f(s)=\sum_{n=1}^\infty a_n n^{-s}\in H^\infty(\Pi)\) with \(\|f\|=1\), the estimate \((\ast)\) \(|a_1|^2+\sum_{n=2}^\infty|a_k|n^{-\sigma}\leq 1\), \(\sigma> \sigma'\), holds for some \(\sigma'\in [1,1.2061[\) (Theorem~3.2). For convex bounded domains, \textbf{r}\((G)\geq 1\) and for convex ones, \textbf{r}\({}_a(G)\geq 1\). The obtained results generalize the classical theorems by Bohr and Rogosinski for the case of Dirichlet series. The isometric Bohr abscissa \textbf{b} was introduced first by \textit{R.~Balasubramanian, B.~Calado} and \textit{H.~Queffélec} [Studia Math. 175, No. 3, 285--304 (2006; Zbl 1110.30001)], where it was shown that \(\log 3/\log 2\leq{\mathbf b}\leq 1.8154\), and \((\ast)\) was proved for some \(1.7286<\sigma<1.7287\).
    0 references
    0 references
    isometric Bohr abscissa
    0 references
    Rogosinski abscissa
    0 references
    Dirichlet series
    0 references