A Radial Basis Function Collocation Method for Time-Dependent Partial Differential Equations: Stability, Accuracy, and Recent Advances
This paper presents an advanced analysis of the radial basis function (RBF) collocation method for solving time-dependent partial differential equations (PDEs). A unified framework is developed that integrates weighted finite difference temporal discretization with global RBF spatial approximation for both smooth and piecewise smooth radial kernels. We provide a rigorous stability analysis of the resulting time-stepping scheme and systematically examine the roles of the shape parameter, polynomial augmentation, and scattered point distribution in controlling accuracy and conditioning. In addition to the classical global collocation formulation, we survey and compare contemporary localized and high-order RBF variants—including localized RBF collocation (LRBF), RBF partition of unity (RBF-PU), ghost-point-enhanced RBF pseudo-spectral (RBF-PS-G), and RBF finite difference (RBF-FD)—that have emerged as the state of the art for large-scale and long-time simulations in 2024–2026. Numerical experiments on benchmark problems including the telegraph and Klein–Gordon equations, a nonlinear viscous Burgers equation, a three-dimensional Helmholtz problem, and comparisons between global RBF, RBF-PU, and LRBF validate the theoretical predictions. The method's simplicity, mesh-free nature, and high-order accuracy make it a competitive and highly adaptable tool for a broad class of PDE problems.
Authors
- Iain S. Burns (ORCID: https://orcid.org/0000-0002-1825-0097)
Institutions
- Islamic Azad University Kerman (IR)
Publication Details
- Journal
- Applied and Computational Mathematics
- Published
- 2026-11-14
- DOI
- https://doi.org/10.11648/j.acm.20261506.12
- Primary Topic
- Numerical methods in engineering
- Type
- article
- Field-Weighted Citation Impact
- 0.00