A Numerical Study of Combining RBF Interpolation and Finite Differences to Approximate Differential Operators
Adrijan Rogan, Andrej Kolar-Požun, Gregor Kosec
TL;DR
This paper compares two RBF-based operator-approximation strategies on a 2D Poisson problem: the standard RBF-FD and a hybrid method that uses RBF interpolation to realize virtual FD stencils. By varying monomial augmentation degree $m$ and virtual-stencil spacing $\delta=\sigma h$ across five-point and nine-point stencils, the study reveals regimes where the hybrid approach can outperform RBF-FD, particularly near the optimal $\sigma$ when $m$ is modest, albeit with higher computational cost. The results highlight that tuning $\sigma$ and $m$ is crucial for maximizing accuracy, and that higher-order RBF-FD generally dominates the hybrid method for large $m$, while the hybrid shines at specific parameter settings for lower $m$. The work underscores the potential of hybrid meshless-FD schemes for adapting existing FD stencils to irregular node layouts, with practical implications for complex geometries and problem-specific discretizations.
Abstract
This paper focuses on RBF-based meshless methods for approximating differential operators, one of the most popular being RBF-FD. Recently, a hybrid approach was introduced that combines RBF interpolation and traditional finite difference stencils. We compare the accuracy of this method and RBF-FD on a two-dimensional Poisson problem for standard five-point and nine-point stencils and different method parameters.
