Recovery of functions from weak data using unsymmetric meshless kernel-based methods (Q2483197)

The unsymmetric Petrov-Galerkin approach is an important method to solve partial differential equations numerically in scientific and engineering applications. Also, the important kernel based methods, such as radial basis function approximations, are highly relevant in this field. In this paper, the kernel methods are analysed for the unsymmetric case, for instance with respect to their stability. In the first place, least squares approximations are used from the spaces spanned by these kernel functions. The paper is completed by a useful error analysis which implies a convergence result and interesting numerical examples.