Piecewise polynomial approximation methods in the theory of Nikol'skiĭ-Besov spaces (Q280904)

This contribution is based on the doctoral dissertation of the author and proposes a new approach to the study of Nikol'skiĭ-Besov spaces. The author introduces and studies dyadic analogs of Nikol'skiĭ-Besov spaces defined by means of piecewise polynomial approximations on the subset of almost dyadic cubes. Then she develops combinatorial algorithms which can be used in applications such as pattern recognition or data compression. Furthermore, from the study of the relation between dyadic and classical Nikol'skiĭ-Besov spaces the author proves several known and new results for classical spaces. In all the paper, the interested reader will find a broad list of novel results corresponding to theorems comparing approximation rates for different norms, decompositions generating piecewise polynomial functions, embedding theorems, interpolation theorems, the order of the dyadic smoothness, a Bernstein-type inequality and a description of diagonal spaces in terms of nonlinear approximations by piecewise polynomial functions.