A well conditioned Method of Fundamental Solutions

Pedro R. S. Antunes

DOI: https://doi.org/10.48550/arXiv.2109.01886

2022-03-22

Abstract:The method of fundamental solutions (MFS) is a numerical method for solving boundary value problems involving linear partial differential equations. It is well known that it can be very effective assuming regularity of the domain and boundary conditions. The main drawback of the MFS is that the matrices involved typically are ill-conditioned and this may prevent to achieve high accuracy. In this work, we propose a new algorithm to remove the ill conditioning of the classical MFS in the context of Laplace equation defined in planar domains. The main idea is to expand the MFS basis functions in terms of harmonic polynomials. Then, using the singular value decomposition and Arnoldi orthogonalization we define well conditioned basis functions spanning the same functional space as the MFS's. Several numerical examples show that this approach is much superior to previous approaches, such as the classical MFS or the MFS-QR.

Numerical Analysis,Optimization and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the ill - conditioning in classical methods, especially when using the Method of Fundamental Solutions (MFS) to solve the Laplace equation. Specifically:

1. **Basic Concepts of MFS**:
- MFS is a numerical method for solving boundary - value problems of linear partial differential equations.
- The solution is approximated by translating the fundamental solutions of the PDE at certain source points, and these source points are located on an allowed source set that does not intersect the solution domain.

2. **Main Problems**:
- **Ill - conditioned Matrix**: One of the main drawbacks of MFS is that the involved matrices are usually ill - conditioned, which may prevent the method from achieving high precision.
- **Source Point Selection**: Another problem is the selection of source points, but this paper mainly focuses on the ill - conditioning problem.

3. **Proposed New Method**:
- The paper proposes a new algorithm, aiming to eliminate the ill - conditioning problem in the classical MFS, especially in the Laplace equation in the planar region.
- The main idea is to expand the MFS basis functions in the form of harmonic polynomials. Then, a set of well - conditioned basis functions are defined by using singular value decomposition (SVD) and Arnoldi orthogonalization, and these basis functions span the same functional space as the original MFS basis functions.

4. **Advantages of the New Method**:
- Numerical experiments show that when applicable, this method is superior to previous methods (such as classical MFS or MFS - QR) in many cases.
- The new method can maintain a low condition number and still maintain high precision when the number of basis functions is increased.

### Formula Representation

- Boundary - value problem of the Laplace equation:
\[
\begin{cases}
\Delta u = 0 & \text{in } \Omega, \\
u = g & \text{on } \partial \Omega,
\end{cases}
\]
where \( g \) is a given function defined on the boundary \( \partial \Omega \).

- Fundamental solution:
\[
\Phi(x)=-\frac{1}{2\pi}\log |x|.
\]

- Direct MFS approximation:
\[
u_{\text{Direct}}^N(x)=\sum_{n = 1}^N c_n^{\text{Direct}}\Phi(x - y_n),
\]
where \( y_n \) are source points, located on an allowed source set \( \hat{\Gamma} \) that does not intersect \( \bar{\Omega} \).

- Condition number of the ill - conditioned matrix:
\[
\text{cond}_2(A_{\text{Direct}})\sim\frac{1}{2\log(R)}R^N.
\]

- Basis function expansion in the new method:
\[
\psi(r,\theta)=M_\infty\cdot F_\infty(r,\theta),
\]
where \( M_\infty \) and \( F_\infty(r,\theta) \) are an infinite - dimensional matrix and a vector function respectively.

Through these improvements, the paper successfully solves the ill - conditioning problem in MFS and improves the precision and stability of the numerical solution.