Kernel theorems for operators on co-orbit spaces associated with localised frames (Q6958884)

Co-orbit spaces are Banach spaces of functions or distributions that satisfy integrability conditions imposed on their generalised wavelet transform. These spaces include well-known examples like modulation spaces and Besov spaces. The use of localised frames in co-orbit spaces is strongly justified. To begin with, they enable sparse atomic decompositions of elements within their corresponding co-orbit spaces. Additionally, by their very nature, the Gram matrices associated with localised frames exhibit sparse structure. Consequently, when combined with Galerkin's method, localised frames are expected to yield sparse matrix representations of bounded linear operators acting on co-orbit spaces. This sparsity is particularly advantageous for the numerical resolution of related operator equations.\N\NKernel theorems offer an effective representation of bounded linear operators, describing their actions as generalised integral operators. In this paper, the authors prove kernel theorems for bounded linear operators acting on co-orbit spaces generated by localised frames. Through this approach, they demonstrate that while the group structure and its associated representation theory offer a very good framework, they are not essential for the proof of kernel theorems.\N\NThe authors characterise the classes of operators whose generalized integral kernels are in the co-orbit spaces of test functions and distributions linked to the tensor product of the localized frames. Additionally, by employing a variant of Schur's test, they derive a characterization of bounded linear operators between particular co-orbit spaces with their kernels in mixed-norm co-orbit spaces. They show that the kernel of any bounded linear operator mapping between co-orbit spaces can be reconstructed from its matrix representation, thereby giving a sparse representation of the kernel itself.