On the choice of basis functions for the meshless radial point interpolation method with small local support domains

Abstract

The behavior of two different types of basis functions for the meshless Radial Point Interpolation Method (RPIM) is investigated in this paper. A 2D test function is interpolated through Gaussian and Wendland basis functions and the approximation errors on the low-order derivatives of the test function are calculated. It is shown that the Gaussian basis function is more appropriate for the interpolation in small support domains whereas Wendland basis function is more accurate for larger support domains.