An approach to wavelet isomorphisms of function spaces via atomic representations (Q1645278): Difference between revisions

The decomposition techniques have become an important tool in the theory of function spaces, and the atomic and wavelet expansions are two widely used examples. These two decomposition techniques have both advantages and disadvantages: the atomic decompositions have high flexibility, but lack unique representation, while the wavelet decompositions have fixed structure and form unconditional bases in many function spaces. Therefore, it could be very useful to build a bridge between these two expansions. It is well known that compactly supported Daubechies wavelets can be directly interpreted as atoms, and hence one automatically obtains an atomic decomposition if there exists a wavelet expansion with compactly supported Daubechies wavelets. \par In this paper, by using Taylor expansion, the authors present a fairly general approach how to construct compactly supported wavelets from an already known atomic decomposition. The considered function space $A(\mathbb{R}^n)$ satisfies $\mathcal{S}(\mathbb{R}^n)\hookrightarrow A(\mathbb{R}^n)\hookrightarrow\mathcal{S}'(\mathbb{R}^n)$ and admits an $L$ atomic decomposition. One key assumption is that the corresponding sequence space $a(\mathbb{R}^n)$ related to the above atomic decomposition should be a $\varkappa$-sequence space for some number $\varkappa\in(0,L)$, a new notion introduced by the authors. Then, it is proved that $A(\mathbb{R}^n)$ also admits a compactly supported Daubechies wavelet decomposition: $f\in \mathcal{S}'(\mathbb{R}^n)$ if and only if it can be represented in terms of $L$-wavelets of the form \[ f=\sum_{m\in\mathbb{Z}^n} \lambda_m \Psi_m+\sum_{G\in G^*}\sum_{j=0}^\infty\sum_{m\in\mathbb{Z}^n} \lambda_m^{j,G}2^{-jn/2}\Psi^j_{G,m} \] with unconditional convergence in $\mathcal{S}'(\mathbb{R}^n)$, and the coefficient sequence belongs to a wavelet sequence space $a^w(\mathbb{R}^n)$. \par This remarkable approach could be of wide applicability since the space $A(\mathbb{R}^n)$ can be chosen as one of many classical function spaces such as (doubling weighted) Besov and Triebel-Lizorkin spaces.