An accurate and asymptotically compatible collocation scheme for nonlocal diffusion problems (Q1669856)

A collocation scheme for the solution of a linear nonlocal diffusion problem with general kernels is developed and analyzed. First, a classic trapezoidal rule which is based on linear interpolation is used in order to approximate the nonlocal diffusion operator. Next, the authors derive a new improved quadrature rule in order to achieve more accurate results and avoid the evaluations of singular integrals. In this way, they construct a collocation method in order to solve the nonlocal diffusion equations. As a result, they obtain a symmetric positive definite matrix having Toeplitz structure. Further, it is shown that the new proposed scheme is of second-order accuracy with respect to the mesh size for a nonlocal problem with a fixed horizon. Additionally, it is shown that it is possible to obtain higher-order accuracy for some commonly used kernels from the literature. Moreover, it is proven that the scheme is asymptotically compatible. This means that the approximate solution of the nonlocal diffusion problem converges to the exact solution of the corresponding local partial differential equation, provided that the horizon and the mesh size tend to zero. Numerical experiments are given to support the theoretical results.