Greedy algorithm for functions with low mixed smoothness
Let \(\{\phi_j\}_{j=1}^\infty\) be a basis of a Banach space \(X\) and for some \(f \in X\) and some real coefficients \(c_j\) let \(f=\sum_{j=1}^\infty c_j\phi_j\). The aim of the \(m\)-term approximation is to find a good approximation to \(f\) by a \(g\) of the form \(g=\sum_k a_k \phi_k\), with at most \(m\) non-zero coefficients. A possible approach is greedy approximation, in which one retains \(m\) terms with the largest values of \(\| c_j \phi_j\| \) in the initial expansion of \(f\) and drops the rest. In this paper, \(X\) is the space \(L_q=L_q(T^d)\) of 1-periodic functions of \(d\) variables on the torus. A special family of bases in \(L_q\), called wavelet-type bases and associated with the dyadic subdivision of the cube \([0,1]^d\), is introduced. If \(\Psi^d\) is such a basis, let \(G_m^q(f, \Psi^d)\) be the approximation to \(f\) obtained by applying the greedy algorithm to its exansion in \(\Psi^d\). It is assumed that \(f\) belongs to the Sobolev class \(MW_p^r\) of functions with a bounded mixed derivative of order \(r\). The focus is on the small values of \(r\) because the case of larger \(r\) has been studied earlier by other authors. Theorem 1. Let \(1 <p,q<\infty\). Then for \((1/p-1/q)_+ < r \leq 1/2-1/q\), \(p,q>2\), \[ \sup_{f \in MW_p^r} \| f-G_m^q(f, \Psi^d)\| _q \asymp m^{-r} (\log m)^\lambda, \quad \lambda={(d-1)r \over 2(r-1/q)}. \] For \((1/p-1/q)_+ < r \leq (\max(2/p, 2/q)-1)/q\), \(q<2\), \[ \sup_{f \in MW_p^r} \| f-G_m^q(f, \Psi^d)\| _q \asymp m^{-r} (\log m)^{(d-1)r}. \] Upper estimates for approximation by \(G_m^q(f, \Psi^d)\) are also found in the case of the Hölder-Nikolski classes \(MH_p^r\) with the small \(r\).
- Nonlinear function approximation: computing smooth solutions with an adaptive greedy algorithm
- Optimality of the greedy algorithm for some function classes
- Convergence of the greedy algorithm for continuous functions
- An algorithm for constrained smoothing functions
- Greedy Algorithms and Rational Approximation in One and Several Variables
- Greedy approximation
- Greedy approximation
- Greedy algorithms for adaptive approximation
- Greedy approximations
- Greedy algorithm for Besov classes
- Continuous algorithms in \(n\)-term approximation and nonlinear widths
- Estimates of diameters of Sobolev classes of small smoothness
- Greedy algorithm for general biorthogonal systems
- Greedy algorithms with regard to multivariate systems with special structure
- scientific article; zbMATH DE number 3743876 (Why is no real title available?)
- scientific article; zbMATH DE number 43504 (Why is no real title available?)
- scientific article; zbMATH DE number 1215245 (Why is no real title available?)
- scientific article; zbMATH DE number 1264507 (Why is no real title available?)
- scientific article; zbMATH DE number 1049042 (Why is no real title available?)
- Hyperbolic wavelet approximation
- Representation and approximation of multivariate functions with mixed smoothness by hyperbolic wavelets
- Universal bases and greedy algorithms for anisotropic function classes
- Greedy algorithms with regard to multivariate systems with special structure
- Nonlinear wavelet approximation of periodic function classes with generalized mixed smoothness
- Greedy algorithm with regard to the needlet system on the sphere
- The greedy ridge algorithm in Gaussian weighted \(L^2\)
- The best \(m\)-term approximations on generalized Besov classes \(M\, B_{q, \theta}^{\Omega}\) with regard to orthogonal dictionaries
- Nonlinear function approximation: computing smooth solutions with an adaptive greedy algorithm
- Nonlinear approximation of function spaces of mixed smoothness
This page was built for publication: Greedy algorithm for functions with low mixed smoothness
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2581445)
