Convergence of series of Fourier coefficients of \(p\)-absolutely continuous functions (Q5932664)

From MaRDI portal
scientific article; zbMATH DE number 1604004
Language Label Description Also known as
English
Convergence of series of Fourier coefficients of \(p\)-absolutely continuous functions
scientific article; zbMATH DE number 1604004

    Statements

    Convergence of series of Fourier coefficients of \(p\)-absolutely continuous functions (English)
    0 references
    12 June 2001
    0 references
    The author extends a theorem of \textit{Stechkin} [Izv. Akad. Nauk SSSR, Ser. Math., 19, 221-246 (1955; Zbl 0065.05405)] and \textit{Bernstein} [On the convergence of trigonometric series, Sochinenia II, 178-183] by giving sufficient conditions for absolute convergence of powers of the Fourier coefficients \(\{a_n, b_n\}\) with weights: \[ \sum_{n = 1}^{\infty} |n|^{\alpha} (|a_n|^{\beta} + |b_n|^{\beta}) M(n). \] The conditions are in terms of best approximation by trigonometric polynomials for functions that are \(p\)-absolutely continuous. The precise statements require the following conditions. Suppose that \(f\) is a bounded, \(2 \pi\)-periodic function, \[ \xi = \{ x_0 < x_1 < \cdots <x_n = x_0 + 2 \pi \} \] is a subdivision of the period, with \(|\xi|= \max\{ x_i - x_{i - 1} :1 \leq i \leq n \}\) being its diameter. The quantity \[ \kappa_{\xi}^p(f) = \left( \sum_{i=1}^n |f(x_i) - f(x_{i - 1})|^p \right)^{1/p} \] is the variational sum of order \(p\) with respect to the subdivision \(\xi\). For \(k \in N\) and \(1 \leq p < \infty\), define \[ \omega_{1 - 1/p} (f, \delta) = \sup \{ \kappa_{\xi}^p(f): |\xi|\leq \delta \}, \] and \[ \omega_{k - 1/p} (f, \delta) = \sup \{ \omega_{1 - 1/p }^p(\Delta_h^{k-1} f(x)): |h|\leq \delta \}, \] where \(k \geq 2\) and \[ \Delta_h^k f(x) = \sum_{i=0}^k (-1)^i \binom ki f(x + ih) \] is the \(k\)-th difference of \(f(x)\) with stepsize \(h>0\). The function \(M \in Q({\delta}), \delta >0\) if \(M\) is a positive, nondecreasing function on \([1, \infty)\) such that \[ M(u +1) - M(u) \leq \delta M(u)/u, u \geq u_0 \geq 1. \] Examples of functions in \(Q(\delta)\) are \(\ln^{\gamma}(u+1), u^{\delta}/\ln^{\gamma}(u+1)\), with \(\gamma \geq 0\) and \(M \equiv 1\). The author's main result is that if \(1<p<\infty, f \in C_p, 0< \beta \leq 2, s = \max\{p,2\}\),\( \alpha - \beta/2 - \beta/s > -1\), if \(M \in Q({\delta})\) for some \(\delta >0\), and \[ \sum_{n = 1}^{\infty} E_n^{\beta}(f)_p n^{\alpha - \beta/2 - \beta/s}M(n) < \infty, \] then \[ \sum_{n = 1}^{\infty} |n|^{\alpha} (|a_n|^{\beta} + |b_n|^{\beta}) M(n) < \infty. \] He also shows that e.g. if \(0 < \beta \leq 1\), and if \(E_n'\) is a sequence with \[ \sum_{n = 1}^{\infty} (E_n')^{\beta} n^{\alpha - \beta/2 - \beta/s}M(n) = \infty, \] there is an \(f \in C_p\) such that \[ E_n(f)_p \leq C_1 E_n'', \;\;\text{for all} \;\;n \in N, \] and \[ \sum_{n = 1}^{\infty} |n|^{\alpha} (|a_n|^{\beta} + |b_n|^{\beta}) M(n) = \infty. \] The result is also true for \(1<\beta \leq 2\) if some restrictions are imposed, e.g. \[ \sum_{k = n+1}^{\infty} E_k''/k \leq C_2 E_n''. . \]
    0 references
    0 references
    Fourier coefficients
    0 references
    \(p\)-absolutely continuous functions
    0 references
    0 references