Nonlinear functionals of wavelet expansions -- adaptive reconstruction and fast evaluation (Q1576614)

This paper is concerned with the efficient evaluation of nonlinear expressions of wavelet expansions obtained through an adaptive process. The authors analyze the local approximation properties of wavelet expansions with respect to lacunary index sets and derive estimates in terms of local best polynomial approximation. Upper bounds are given in terms of local Besov norms in combination with a certain basic adaptive refinement scheme. These results are transferred to the reconstruction of compositions of functions and to the analysis of its complexity. Numerically implementable reconstruction schemes that preserve the desired local approximation properties are studied. This leads in combination with two local basis transformations to a concrete algorithm for the approximate evaluation of wavelet coefficients of compositions of functions. The asymptotically optimal efficiency of this scheme is investigated. A simple numerical example is given.