Algorithms for cardinal interpolation using box splines and radial basis functions.
The paper deals with the computation of suitably chosen parameters of a biparabolic spline (ot the tensor product type) on a rectangular domain. Some possibilities of choosing such local parameters (concentrated, dispersed parameters) are discussed. The algorithms for computation of dispersed parameters (using the first derivative representation) and concentraced parameters (using the second derivative representation) are given. Both these algorithms repeatedly use the one-dimensional algorithms....
The paper discusses new cubature formulas for classical integral operators of mathematical physics based on the «approximate approximation» of the density with Gaussian and related functions. We derive formulas for the cubature of harmonic, elastic and diffraction potentials approximating with high order in some range relevant for numerical computations. We prove error estimates and provide numerical results for the Newton potential.
The paper deals with approximation of locally Lipschitz functionals. A concept of approximation, based on the idea of graph approximation of the generalized gradient, is discussed and the existence of such approximations for locally Lipschitz functionals, defined on open domains in , is proved. Subsequently, the procedure of a smooth normal approximation of the class of regular sets (containing e.g. convex and/or epi-Lipschitz sets) is presented.
We discuss best N-term approximation spaces for one-electron wavefunctions and reduced density matrices ρ emerging from Hartree-Fock and density functional theory. The approximation spaces for anisotropic wavelet tensor product bases have been recently characterized by Nitsche in terms of tensor product Besov spaces. We have used the norm equivalence of these spaces to weighted spaces of wavelet coefficients to proof that both and ρ are in for all with . Our proof is based on the...