A local target specific quadrature by expansion method for evaluation of layer potentials in 3D

Michael Siegel,Anna-Karin Tornberg
DOI: https://doi.org/10.1016/j.jcp.2018.03.006
2018-03-06
Abstract:Accurate evaluation of layer potentials is crucial when boundary integral equation methods are used to solve partial differential equations. Quadrature by expansion (QBX) is a recently introduced method that can offer high accuracy for singular and nearly singular integrals, using truncated expansions to locally represent the potential. The QBX method is typically based on a spherical harmonics expansion which when truncated at order $p$ has $O(p^2)$ terms. This expansion can equivalently be written with $p$ terms, however paying the price that the expansion coefficients will depend on the evaluation/target point. Based on this observation, we develop a target specific QBX method, and apply it to Laplace's equation on multiply connected domains. The method is local in that the QBX expansions only involve information from a neighborhood of the target point. An analysis of the truncation error in the QBX expansions is presented, practical parameter choices are discussed and the method is validated and tested on various problems.
Numerical Analysis
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in three - dimensional multiply - connected domains, how to accurately evaluate layer potentials, especially singular and near - singular integrals. Specifically, the author proposes a local target - specific Quadrature by Expansion (QBX) method to calculate the layer potentials of the Laplace equation with high precision. ### Problem Background When using the boundary integral equation method to solve partial differential equations, accurate evaluation of layer potentials is crucial. However, when dealing with points close to the boundary or problems in multiply - connected domains, singular and near - singular integrals will occur, which makes numerical solution complex and difficult to calculate precisely. Traditional quadrature methods may produce large errors in such cases. ### Core Problems of the Paper 1. **Accurate Evaluation of Singular and Near - Singular Integrals**: Especially in three - dimensional multiply - connected domains, when the evaluation point is close to the boundary, how to calculate the layer potentials efficiently and accurately. 2. **Improving Computational Efficiency**: Reducing the computational complexity so that the method can be applied to large - scale problems, such as electromagnetic scattering problems involving a large number of geometries or particulate Stokes flow problems. ### Proposed Method The paper proposes a local target - specific QBX method. The main features of this method include: - **Locality**: It only involves local information near the target point, reducing the complexity of global computation. - **High Precision**: By truncating the expansion to represent the layer potential and using spherical harmonic function expansion or other forms of expansion, high precision is ensured. - **Adaptability**: It can be applied to smooth boundaries of arbitrary shapes without pre - computation and is suitable for time - dependent geometric problems. ### Key Innovation Points of the Method - **Target - Specific QBX Expansion**: Different from the traditional global QBX method, this method uses target - specific expansion, reducing the number of required terms and improving computational efficiency. - **Error Analysis**: A detailed analysis of truncation error and coefficient error is carried out, providing a theoretical basis for parameter selection. - **Complexity Optimization**: By combining local correction and the fast multipole method (FMM), an O(N) or O(N log N) complexity is achieved. ### Application Scenarios This method is especially suitable for problems that require frequent evaluation of near - singular integrals, such as: - Electromagnetic scattering problems - Particulate Stokes flow problems - Blood flow simulation Through these improvements, this method can handle the layer potential calculation problems in complex three - dimensional multiply - connected domains more efficiently and accurately.