Paper: Moving Least Square Reproducing Kernel Method Part I: Methodology and Convergence

W. K. Liu, S. Li and T. Belytschko

Accepted for publication in "Computer Methods in Applied Mechanics and Engineering"


Abstract

This paper reformulates the moving least square interpolation scheme, in a framework of the so-called moving least square kernel (MLSK) representation. In this study, the procedure of constructing moving least square interpolation function is facilitated by using the notion of reproducing kernel formulation, which is a generalization of the early discrete approach. Here, the concept of reproducing kernel integral representation is particularly emphasized, such that the method not only shares its root with finite element method, but also is lined with the multiresolution analysis in general. The ``shape functions'' constructed by this method form a complete basis of finite dimensional Hilbert space; moreover, the reproducing formula is not only able to reproduce any m-th order polynomial exactly, but also able to reproduce any smooth function accurately in a global least square sense. By tuning the dilation parameter to a proper length, the subspace method that is equipped with this particular interpolant function can achieve an optimal accuracy, in a fixed particle distribution, to the response corresponding to the source of a certain wave number or frequency band. An interpolation error estimate is given to assess the convergence rate of the numerical computation. It is shown that for sufficiently smooth function the interpolant expansion in terms of sampled values will converge to the original function in Sobolev norm. As a meshless method, the convergence rate is measured by a new control variable --- dilation parameter ha of window function, instead of the mesh size h as usually done in the finite element analysis. To illustrate the technical procedure, convergence has been shown for the numerical solution of second order elliptic differential equations in a Galerkin procedure invoked with this interpolant. As an numerical example, a two point boundary problem is solved by using the method, and a very fast convergence rate is observed with respect to various norms.

This paper is available in postcript form. (66 pages, ~1.1MB)
This paper is available in Adobe Acrobat form
 

Professor Wing Kam Liu

Send email: w-liu@nwu.edu