An extremum problem for distribution functions (Q1272300)

From MaRDI portal
scientific article
Language Label Description Also known as
English
An extremum problem for distribution functions
scientific article

    Statements

    An extremum problem for distribution functions (English)
    0 references
    0 references
    0 references
    16 August 2000
    0 references
    Let us consider the functions of the form \[ f( t)= \sum_{\nu \in Z} \widehat {f}_\nu e^{i\nu t},\quad \sum_{\nu \in Z} |\widehat {f}_\nu|< + \infty . \tag{1} \] The following result is proved. Theorem 1. For any \(n \in N\) and any nonnegative function \(f \not\equiv 0\) of the form \((1)\) whose Fourier coefficients satisfy the conditions \[ \widehat f_\nu = {1\over 2\pi} \int_{0}^{2\pi} f(t)e^{i\nu t} dt \leq 0,\quad |\nu|\geq n,\quad \nu \in Z, \] the following inequality is valid: \[ \text{meas}\{ t \in [0,2\pi):f(t) > 0\} \geq {2\pi\over n}. \] The assertion of this theorem is sharp for any \(n \in N\). This is shown by the example of the \(2\pi - \)periodic function: \[ f(t) \equiv f_n (t) = \begin{cases} |\sin nt |,& |t|\leq {\pi\over n},\\ 0,& {\pi\over n} \leq |t|\leq \pi. \end{cases} . \]
    0 references
    0 references
    0 references
    0 references
    0 references
    extremum problem
    0 references
    distribution function
    0 references
    Fourier coefficients
    0 references
    absolute convergence
    0 references