Tristan Goodwill,Charles L. Epstein
Abstract:We introduce a new class of computationally tractable scattering problems in unbounded domains, which we call decomposable problems. In these decomposable problems, the computational domain can be split into a finite collection of subdomains in which the scatterer has a "simple" structure. A subdomain is simple if the domain Green's function for this subdomain is either available analytically or can be computed numerically with arbitrary accuracy by a tractable method. These domain Green's functions are then used to reformulate the scattering problem as a system of boundary integral equations on the union of the subdomain boundaries. This reformulation gives a practical numerical method, as the resulting integral equations can then be solved, to any desired degree of accuracy, by using coordinate complexification over a finite interval, and standard discretization techniques.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the scattering problem of **unbounded interfaces**, especially the wave - scattering phenomenon in waveguide systems with complex geometric structures and infinite boundaries. Specifically, the authors focus on how to solve these scattering problems numerically in an efficient and accurate manner.
### Problem Background
In many practical applications, such as in the fields of optics, acoustics, and electromagnetics, it is very important to study how waves scatter on unbounded interfaces. However, due to the complexity of these problems and the particularity of the boundary conditions, traditional numerical methods are often difficult to handle such problems, especially when it comes to infinite boundaries or non - compact scatterers.
### Core Contributions of the Paper
1. **Introducing the Concept of Decomposable Problems**:
- The authors propose a new class of computationally feasible scattering problems, called "decomposable problems". In these problems, the computational domain can be divided into several sub - domains, and within each sub - domain, the scatterer has a simple structure.
- The simplicity of the sub - domain means that its Green's function can either be directly obtained analytically or can be calculated to any desired accuracy by effective numerical methods.
2. **Reconstructing the Scattering Problem Using Green's Function**:
- By using the Green's functions of these sub - domains, the authors reformulate the original scattering problem as a series of boundary integral equations, which can be solved on the union of the sub - domain boundaries.
- This reformulation makes numerical solution feasible because the resulting integral equations can be solved to any desired accuracy within a finite interval by coordinate complexification and standard discretization techniques.
3. **Specific Numerical Methods**:
- The authors describe in detail how to construct these Green's functions and show how to use these Green's functions to set up and solve the boundary integral equations.
- They also provide a series of numerical experiments to verify the effectiveness and accuracy of the proposed method.
### Summary of Mathematical Formulas
- Basic form of the scattering problem:
\[
(L + k^2)u = f
\]
where \( L=\Delta + q(x)\), and \( f\) is the source term defined by the "incident data".
- Representation of Green's function:
\[
G_k(x; y)=\frac{i}{4\pi}H_0^{(1)}(k|x - y|)
\]
where \(H_0^{(1)}\) is the first - kind Hankel function.
- Representation of the boundary integral equation:
\[
\begin{pmatrix}
I & K_1 \\
K_3 & I + K_4
\end{pmatrix}
\begin{pmatrix}
\sigma \\
\tau
\end{pmatrix}
=
\begin{pmatrix}
-v_{\text{inc}} \\
\partial_n v_{\text{inc}}
\end{pmatrix}
\]
### Conclusion
By this method, the authors can not only effectively solve the scattering problems of unbounded interfaces with complex geometric structures, but also ensure the uniqueness and physical meaning of the solutions. In addition, this method has been verified in a variety of different types of waveguide systems, demonstrating its wide applicability and reliability.