On Sobolev and Besov Spaces With Hybrid Regularity

Helmut Harbrecht,Remo von Rickenbach
2024-11-08
Abstract:The present article is concerned with the nonlinear approximation of functions in the Sobolev space H^q with respect to a tensor-product, or hyperbolic wavelet basis on the unit n-cube. Here, q is a real number, which is not necessarily positive. We derive Jackson and Bernstein inequalities to obtain that the approximation classes contain Besov spaces of hybrid regularity. Especially, we show that all functions that can be approximated by classical wavelets are also approximable by tensor-product wavelets at least at the same rate. In particular, this implies that for nonnegative regularity, the classical Besov spaces of regularity q+sn, integrability and weak index t, with 1/t = s + 1/2, are included in the Besov spaces of hybrid regularity with isotropic regularity q and additional mixed regularity s.
Functional Analysis
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the nonlinear approximation problem of functions in Sobolev space \(H^q\), especially the nonlinear approximation on the unit n - dimensional cube using tensor products or hyperbolic wavelet basis. Specifically, the author focuses on how to approximate functions with mixed regularity through these bases, and derives Jackson's inequality and Bernstein's inequality to prove that functions in Besov spaces containing mixed regularity can be approximated by these bases at at least the same rate. ### Specific problem description: 1. **Nonlinear approximation**: Research how to use tensor products or hyperbolic wavelet basis to perform nonlinear approximation of functions in Sobolev space \(H^q\). 2. **Mixed regularity**: Explore the approximation properties of functions with mixed regularity (i.e., functions with different regularities in different directions). 3. **Jackson and Bernstein inequalities**: By deriving these inequalities, prove that functions in Besov spaces with mixed regularity can be effectively approximated using tensor products or hyperbolic wavelet basis. 4. **Compare the approximation capabilities of different bases**: Prove that for the case of non - negative regularity, the approximation capabilities of the classical wavelet basis and the tensor - product wavelet basis are the same. ### Main objectives: - **Theoretical framework**: Establish a theoretical framework for analyzing and comparing the performance of different bases (such as tensor - product wavelet basis and classical wavelet basis) in nonlinear approximation. - **Regularity conditions**: Determine which regularity - conditioned functions can be effectively approximated by these bases. - **Scope of application**: Explore the significance of these results in practical applications, especially in numerical analysis and partial differential equation solving. ### Key results: - **Jackson and Bernstein inequalities**: Through these inequalities, it is proved that functions in Besov spaces with mixed regularity can be effectively approximated using tensor - product wavelet basis. - **Inclusion relations**: Prove that all functions that can be approximated by the classical wavelet basis can also be approximated by the tensor - product wavelet basis at at least the same rate. - **Non - negative regularity**: In particular, for the case of non - negative regularity, the classical Besov space \(B^{q + s}_{\tau,\tau}\) is included in the mixed regularity Besov space \(B^{q,s}_{\tau,\tau}\). ### Application prospects: - **Numerical analysis**: These results can be used to improve numerical methods, especially when dealing with functions with complex regularity. - **Partial differential equations**: When solving partial differential equations, choosing appropriate basis functions can significantly improve computational efficiency and accuracy. - **Signal processing**: In the field of signal processing, these results can be used to design more efficient signal compression and recovery algorithms. Through these studies, the author provides new perspectives and tools for understanding and optimizing the nonlinear approximation of high - dimensional functions.