Ubiquitous evaluation of layer potentials using quadrature by kernel-independent expansion (Q1647663)

This paper deals with Dirichlet boundary-value problems \[ \mathcal{L}u=0 \text{ in } \Omega, \; u=f \text{ on } \Gamma \tag{1} \] where \(\Omega\) is a simply-connected interior domain with smooth boundary \(\Gamma\), and \[ \mathcal{L}u=\begin{cases} \Delta u, & \text{Laplace;} \\ (\Delta-\lambda^2)u, & \text{Yukawa;} \\ (\Delta+\omega^2)u, & \text{Helmholtz} \; (Im\,\omega\geq 0); \\ \Delta u-\nabla p, & \text{Stokes;} \\ \Delta u+\frac{1}{1-2\nu}\nabla\nabla\cdot u, & \text{Elastostatic.} \end{cases}\tag{2} \] The solution of problems (1)-(2) for \(x\in\Omega\) is represented by the double-layer potentials \[ u(\mathbf{x})=\int_{\Gamma}\frac{\partial\Phi(\mathbf{x,y})}{\partial \mathbf{n_y}}\phi(\mathbf{y})ds(\mathbf{y}), \tag{3} \] where \(\Phi\) is the fundamental solution for the operator \(\mathcal{L}\), and \(\phi\) is an unknown density which satisfies some boundary integral equation of the second kind. The authors introduce a quadrature scheme for the \textit{ubiquitous} high-order accurate evaluation of singular layer potentials associated with general elliptic PDEs, i.e., a scheme that yields high accuracy at all distances to the domain boundary as well as on the boundary itself. Relying solely on point evaluations of the underlying kernel, this scheme is essentially PDE-independent; in particular, no analytic expansion nor addition theorem is required. Moreover, it applies to boundary integrals with singular, weakly singular, and hypersingular kernels. This work builds upon quadrature by expansion, which approximates the potential by an analytic expansion in the neighborhood of each expansion center. In contrast, they use a sum of fundamental solutions lying on a ring enclosing the neighborhood, and solve a small dense linear system for their coefficients to match the potential on a smaller concentric ring. The method is tested with Laplace, Helmholtz, Yukawa, Stokes, and Navier (elastostatic) kernels in two dimensions (2D) using adaptive, panel-based boundary quadratures on smooth and corner domains. Advantages of the algorithm include its relative simplicity of implementation, immediate extension to new kernels, dimension-independence (allowing simple generalization to 3D), and compatibility with fast algorithms such as the kernel-independent FMM.