Combinations of multivariate averages (Q707209)

The article is devoted to investigation of the rate of approximation of a function \(f\) by some combinations of averages on the spheres. One of this averages is given by \[ V_{l,t}f\left(x\right)= {-2\over{\binom{2l}{l}}} \sum_{j=1}^l\left(-1\right)^j \binom{2l}{l-j}V_{jt}f\left(x\right). \] Here \[ V_tf\left(x\right)={1\over{m\left(t\right)}}\int_{\left\{y\in{\mathbb R}^d:\,\left| x-y\right| =t\right\}} f\left(y\right)\,d\sigma\left(y\right),\quad V_t1=1,\quad x\in{\mathbb R}^d, \] where \(d\sigma\left(y\right)\) is a measure invariant under rotations about \(x\). It is shown that the mentioned rate of approximation is equivalent to some \(K\)-functional. In particular, the following statement is obtained. \textbf{Theorem.} For \(d>1\), \(l=1,2,\dots,t>0\) and \(1\leq p\leq\infty\) we have \[ \left\| V_{l,t}f-f\right\| _{L_p\left( {\mathbb R}^d\right)}\approx \inf_g\left(\left\| f-g\right\| _p+t^{2l}\left\| \Delta^lg\right\| _p\right)\equiv K_l\left(f,\Delta,t^{2l}\right)_p \] (here \(\Delta^l\) denotes the \(l\)-th degree of the Laplacian). The reader can find several other interesting results presented in the article, like the following Bernstein-type inequality that may be very useful in the investigations. Theorem. Suppose \(f\in L_p\left({\mathbb R}^d\right)\), \(1\leq p\leq\infty\) and \(\text{supp}\, \hat f\subset\left\{x:\,\left| x\right| \leq R\right\}\). Then \(\Delta^lf\) exists in \(L_p\) and \[ \left\| \Delta^lf\right\| _p\leq CR^{2l}\left\| f\right\| _p \] with \(C\) independent of \(R\) and \(p\). A Jackson-type result for approximation in \(L_p\left({\mathbb R}^d\right)\) by functions, whose Fourier transform is supported in \(\left\{x:\left| x\right| \leq\lambda\right\}\), is also obtained: \[ E_\lambda\left(f\right)\leq K_l\left(f,\Delta,\lambda^{-2l}\right)_p. \] The article contains several deep and useful results and should be very interesting for the specialists in approximation theory and other fields of analysis.