FUNDAMENTALNAYA I PRIKLADNAYA MATEMATIKA

(FUNDAMENTAL AND APPLIED MATHEMATICS)

1998, VOLUME 4, NUMBER 1, PAGES 245-302

An estimate of the minimum of the absolute value of trigonometric polynomials with random coefficients

A. G. Karapetian

Abstract

View as HTML     View as gif image    View as LaTeX source

In this paper the random trigonometric polynomial $T(x)=\sum\limits_{j=0}^{n-1}\xi_j \exp (ijx)$ is studied, where $\xi,\xi_j$ are real independent equally distributed random variables with zero mathematical expectations, positive second and finite third absolute moments.
\begin{theorem}
For any $\varepsilon\in (0,1)$ and $n>(C(\xi))^{7654/\varepsilon^3}$
$$
\mathsf{Pr} \biggl(\min_{x\in\ttt}
\biggl| \sum_{j=0}^{n-1}\xi_j \exp(ijx) \biggr| >
n^{-\frac{1}{2}+\varepsilon}\biggr) \leq \frac{1}{n^{\varepsilon^2/62}},
$$

where $C(\xi)$ is defined in the paper.
\end{theorem}
In the proof of the theorem we use the method of normal degree and establish the estimates for probabilities of events $E_k$, $k\in\mathbb{N}$, $0<k<\frac{k_0}{2}$, and their pairwise intersections. The events $E_k$ are defined by random vectors $X$:
\begin{multline*}
X=(\Re T(x_k),\ldots,\Re (T^{(r-1)}(x_k)/(in)^{r-1}),\\
\Im T(x_k),\ldots,\Im (T^{(r-1)}(x_k)/(in)^{r-1})),
\end{multline*}
where $r$ is chosen as a natural number, such that $\frac{10}{\varepsilon}<r<\frac{11}{\varepsilon}$ for given $\varepsilon$ and $x_k=\frac{2\pi k}{k_0}$, where $k_0$ is the greatest prime number, not greater then $n^{1-\frac{\varepsilon}{20}}$.

To find these estimates first of all we obtain inequalities for polynomials and by these inequalities we establish the properties of characteristic functions of random vectors $X$ and their pairwise unions.

All articles are published in Russian.

Main page Editorial board
Instructions to authors Contents of the journal

Location: http://mech.math.msu.su/~fpm/eng/98/981/98120t.htm
Last modified: April 8, 1998