Optimal sampling recovery of mixed order Sobolev embeddings via discrete {L}ittlewood--{P}aley type characterizations (Q1677612)

The authors study the problem of the optimal sampling recovery in classes of multivariate functions. The \(L_{q}\)-approximation of multivariate functions \(f\) with \(L_{p}\)-bounded mixed derivative (difference) is considered. For this reason it is given the construction of the univariate interpolation operators and the corresponding kernels, and are defined and discussed several characterizations of Besov-Triebel-Lizorkin spaces. Furthermore, tools for estimating the norms of superpositions of dyadic trigonometric polynomials in Besov-Triebel-Lizorkin spaces are provided, and it is given a proof for the sampling representations, which are used to analyze Smolyak's algorithm. Finally, the obtained results are compared with linear and Gelfand \(n\)-widths. The Appendix contains some basic facts and known results on maximal inequalities, Fourier transform, Fourier series, exponential sums and \(n\)-widths.