Estimating the Condition Number for Multivariate Interpolation Problems

P. Binev,K. Jetter
DOI: https://doi.org/10.1007/978-3-0348-8619-2_3
1992-01-01
Abstract:We estimate the condition number in multivariate interpolation problems where the interpolation space stems from shifted versions of a single basis function φ. The estimate depends on the properties of the Fourier transform φ∧. In case the interpolation points are a subset of a regular grid, the bound is based on properties of the symbol of φ (which is a periodization of φ∧). For scattered data interpolation, we assume that φ is a positive definite function where the Fourier transform can be properly bounded. As examples, we consider interpolation by box splines or by certain radial basis functions.
What problem does this paper attempt to address?