RUS  ENG
Full version
JOURNALS // Mathematical Physics and Computer Simulation // Archive

Mathematical Physics and Computer Simulation, 2019 Volume 22, Issue 1, Pages 24–34 (Mi vvgum247)

Mathematics and mechanics

On approximation of the functions of two variables by some Fourier integrals

Yu. Kh. Khasanov

Russian-Tajik Slavonic University

Abstract: This paper we studies some issues on the deviation of the functions of two variables $f(x,y)$ defined on the whole two-dimensional space from integral mean values of their Fourier transforms in the metric of the space $L_p (R^2 )\,\,\,(1\leq p < \infty)$.
Let $L_p (R^2 )\,\,\,(1\leq p < \infty)$ stand for the space of measurable functions $f(x,y)$ such that
$$ \|f(x,y)\|_{L_p}=\left\{\int\limits_{-\infty}^\infty\int\limits_{-\infty}^\infty|f(x,y)|^pdxdy\right\}^{\frac{1}{p}}<\infty\,\,\,(1\leq p<\infty), $$

$$ \|f(x,y)\|_{L_\infty}=vrai\sup_{x,y}|f(x,y)|<\infty, $$
and almost everywhere there exists the Fourier transform
$$ F(t,z)=\frac{1}{2\pi}\int\limits_{-\infty}^\infty\int\limits_{-\infty}^\infty f(u,v) \exp(-i(tu+zv))dudv, $$
where
$$ F(t,z)\in L_q(R^2)\,\,\,(\frac{1}{p}+\frac{1}{q}=1). $$
For any $\sigma>0$ we consider
$$ S_{\sigma,\sigma}(f;x,y)=\int\limits_{-\sigma}^\sigma\int\limits_{-\sigma}^\sigma F(t,z) \exp(i(tx+zy))dtdz= $$

$$ =\int\limits_0^\sigma\left\{\int\limits_{-u}^uA(t,u)dt+\int\limits_{-u}^uA(t,-u)dt+\int\limits_{-u}^uA(u,z)dz+\int\limits_{-u}^uA(-u,z)dz\right\}du= $$

$$ =\int\limits_0^\sigma S_{u,u}^*(f;x,y)du, $$
where $A(t,z)=F(t,z)\exp(i(tx+zy))$.
This paper estimates the value
$$ R_{\sigma,r}(f)_{L_p}=\|f(x,y)-U_{\sigma,r}(f;x,y)\|_{L_p}, $$
where
$$ U_{\sigma,r}(f;x,y)=\int\limits_0^\sigma \left(1-\frac{u^r}{\sigma^r}\right) S_{u,u}^*(f;x,y)du. $$

Theorem 1. If $f(x,y)\in L_p(R^2)\,\,\,(1<p\leq 2)$, then the following bound is valid
$$ R_{\sigma,r}(f)_{L_p}\leq C_{p,r}\left\{\omega_r^{(1)}(f;\frac{1}{\sigma})_{L_p}+\omega_r^{(2)}(f;\frac{1}{\sigma})_{L_p}\right\}, $$
where
$$ \omega_r^{(1)}(f;u)_{L_p}=\sup_{|h|\leq u}\|\Delta_{x,h}^rf\|_{L_p}= \sup_{|h|\leq u}\left\|\sum_{\nu=0}^r(-1)^{r-\nu}(_\nu^r)f(x+\nu h,y)\right\|_{L_p}, $$

$$ \omega_r^{(2)}(f;u)_{L_p}=\sup_{|h|\leq u}\|\Delta_{h,y}^rf\|_{L_p}= \sup_{|h|\leq u}\left\|\sum_{\nu=0}^r(-1)^{r-\nu}(_\nu^r)f(x,y+\nu h)\right\|_{L_p}, $$
$C_{p,r}$ is a constant value that depends only on $p$ and $r$.
Theorem 2. Under the assumptions of Theorem 1 with $1<p\leq 2$ the following bound is valid
$$ \omega_r^{(\nu)}(f;\frac{1}{\sigma})_{L_p}\leq M_{p,r}R_{\sigma,r}(f)_{L_p}\,\,\,(\nu=1,2), $$
where the constant $M_{p,r}$ depends only on $p$ and $r$.

Keywords: function of two variables, Fourier series, Fourier transformation, partial sums of Fourier series, integral mean values, entire function of finite order, best approximation, modulus of continuity.

UDC: 517.518.68
BBK: 22.161.5

DOI: 10.15688/mpcm.jvolsu.2019.1.3



© Steklov Math. Inst. of RAS, 2024