A classification of anisotropic Besov spaces (Q2197951)

The paper is devoted to the study for the classification problem of expansive dilation matrices which induce different anisotropic Besov spaces, and the dependence of the scale of anisotropic Besov spaces on the underlying matrix. To this end, the authors introduce the equivalent relation \(A \sim_{\dot B} B\) for two expansive matrices \(A\) and \(B\), which means that their related homogeneous anisotropic Besov spaces, \(\dot B^{\alpha}_{p,q}(A)\) and \(\dot B^{\alpha}_{p,q}(B)\), coincide for all \(\alpha\in\mathbb{R}\) and \(p,\,q\in(0,\infty]\); the relation \(A \sim_B B\) for inhomogeneous cases is defined analogously. The first step is to construct the topological isomorphisms from \(\dot B^{\alpha}_{p,q}(A)\) and \(B^{\alpha}_{p,q}(A)\) to some decomposition spaces generated by the covering induced by \(A\), which allows to transfer problems from the function spaces to the combinatorial properties of the underlying coverings. In particular, the question whether two matrices \(A\) and \(B\) yield the same anisotropic Besov spaces is transferred to a question about the relationship between their associated quasi-norms. It is then proved that \(A \sim_{\dot B} B\) if and only if there exists \((p,q)\neq (2,2)\) and \(\alpha\in\mathbb{R}\) such that \(\dot B^{\alpha}_{p,q}(A)=\dot B^{\alpha}_{p,q}(B)\), and \(A \sim_{B} B\) if and only if there exists \((p,q)\neq (2,2)\) and \(\alpha\in\mathbb{R}\) such that \(B^{\alpha}_{p,q}(A)=B^{\alpha}_{p,q}(B)\). In particular, \(A \sim_{\dot B} B\) implies \(A \sim_{B} B\), but the converse of this statement is generally false. Thus, two matrices \(A\), \(B\) that induce the same scale of homogeneous Besov spaces also induce the same scale of inhomogeneous spaces. Applying the above result, the authors further show that, for each expansive matrix \(A\), there exists an expansive matrix \(A'\) (called expansive normal form of \(A\)) such that \(A'\) has only positive eigenvalues, \(|\mathrm{det}(A )|=2\), and the norms induced by the two matrices are equivalent. Finally, via proving any two matrices in expansive normal form induce equivalent quasi-norms if and only if they are equal, they provide an algorithm to decide whether two expansive matrices \(A\) and \(B\) induce the same Besov spaces.