Local-basis Difference Potentials Method for elliptic PDEs in complex geometry

TL;DR

This paper introduces a high-order finite difference method using local basis functions within the Difference Potentials framework to efficiently solve elliptic PDEs in complex geometries, accommodating intricate shapes and boundary conditions.

Contribution

The paper presents a novel local basis function approach in the Difference Potentials method, enhancing flexibility and accuracy for elliptic PDEs in complex geometries.

Findings

Achieves high-order accuracy in complex geometries.

Demonstrates flexibility in handling various boundary conditions.

Provides convergence proofs both theoretically and numerically.

Abstract

We develop efficient and high-order accurate finite difference methods for elliptic partial differential equations in complex geometry in the Difference Potentials framework. The main novelty of the developed schemes is the use of local basis functions defined at near-boundary grid points. The use of local basis functions allow unified numerical treatment of (i) explicitly and implicitly defined geometry; (ii) geometry of more complicated shapes, such as those with corners, multi-connected domain, etc; and (iii) different types of boundary conditions. This geometrically flexible approach is complementary to the classical difference potentials method using global basis functions, especially in the case where a large number of global basis functions are needed to resolve the boundary, or where the optimal global basis functions are difficult to obtain. Fast Poisson solvers based on FFT are employed for standard centered finite difference stencils regardless of the designed order of accuracy. Proofs of convergence of difference potentials in maximum norm are outlined both theoretically and numerically.