A finite point method for compressible flow

Abstract

A weighted least squares finite point method for compressible flow is formulated. Starting from a global cloud of points, local clouds are constructed using a Delaunay technique with a series of tests for the quality of the resulting approximations. The approximation factors for the gradient and the Laplacian of the resulting local clouds are used to derive an edge-based solver that works with approximate Riemann solvers. The results obtained show accuracy comparable to equivalent mesh-based finite volume or finite element techniques, making the present finite point method competitive.