Solving the Scattering Problem for Open Wave-Guide Networks, I Fundamental Solutions and Integral Equations

Technical Summary: Solving the Scattering Problem for Open Wave-Guide Networks, I

Problem Statement
This paper addresses the scalar scattering problem for open, dielectric wave-guide networks in two dimensions. Specifically, it focuses on the simplest non-trivial configuration: two semi-infinite dielectric channels (wave-guides) meeting along a straight interface orthogonal to their propagation directions. The system is modeled by the time-harmonic Schrödinger equation $(\Delta + \omega^2 \epsilon(x))u = 0$, where the permittivity $\epsilon(x)$ is constant outside a compact set and within tubular neighborhoods of the semi-infinite rays.

The core difficulty lies in the non-compact nature of the scatterer. Unlike bounded scatterers where the Sommerfeld radiation condition suffices to define outgoing solutions, wave-guide networks require more complex radiation conditions that account for guided modes traveling along the channels and radiating fields in the exterior. Furthermore, standard limiting absorption principles face challenges because the error terms introduced by cutting off incoming fields often fail to belong to the necessary weighted $L^2$ spaces. The paper aims to reformulate this scattering problem as a transmission problem across an artificial boundary (the interface $x_1=0$) and solve it via integral equations.

Methodology
The authors employ a layer potential approach based on the construction of explicit outgoing fundamental solutions for the translation-invariant operators defining the two ends of the network.

1. Construction of Fundamental Solutions:
   The fundamental solutions $E_{l,r}(x; y)$ for the operators $\Delta + k_1^2 + q_{l,r}(x_2)$ are constructed using the Fourier transform in the $x_1$-variable. The solution is decomposed into a free-space term (the Hankel function $H_0^{(1)}$) and a correction term $w_{l,r}$. The correction term is found by solving an ordinary differential equation (ODE) in the $x_2$ variable for each frequency $\xi$, utilizing the classical theory of ODEs and the limiting absorption principle ($\delta \to 0^+$).

   • The continuous spectrum contribution is represented via a contour integral $\Gamma_\nu^+$ in the complex $\xi$-plane, deformed to avoid the roots of the Wronskian.
   • The discrete spectrum (guided modes) is captured by the residues at the poles $\pm \xi_n$, resulting in terms proportional to $v_n(x_2)v_n(y_2)e^{i\xi_n(x_1-y_1)}$.

2. Layer Potential Representation:
   The scattered fields $u_l$ (for $x_1 < 0$) and $u_r$ (for $x_1 > 0$) are represented as combinations of single and double layer potentials using these fundamental solutions:
   $$u_{l,r} = -E_{l,r}' \sigma + E_{l,r} \tau$$
   where $\sigma$ and $\tau$ are unknown source densities on the interface $x_1=0$.

3. Integral Equation Formulation:
   By enforcing the transmission boundary conditions (continuity of the field and its normal derivative, adjusted for incoming data), the authors derive a system of integral equations for $(\sigma, \tau)$ on the line $x_1=0$. The system takes the form:
   $$\begin{pmatrix} I & D \\ C & I \end{pmatrix} \begin{pmatrix} \sigma \\ \tau \end{pmatrix} = \begin{pmatrix} g \\ h \end{pmatrix}$$
   Here, $D$ and $C$ are integral operators with kernels derived from the correction terms $w_{l,r}$ and their derivatives. Notably, the singularities of these kernels are milder than standard layer potentials (behaving like $|x-y|^2 \log|x-y|$ rather than $|x-y|^{-1}$).

4. Functional Analytic Setting:
   The analysis is conducted in Banach spaces of continuous functions with specific decay rates, $C_\alpha(\mathbb{R})$, defined by the norm $|f|_\alpha = \sup (1+|x|)^\alpha |f(x)|$. The authors establish that the operators $C$ and $D$ map between these spaces and that the composition $DC$ is compact.

Key Contributions and Results

• Explicit Fundamental Solutions: The paper provides a constructive method for generating outgoing fundamental solutions for perturbed wave-guide operators, separating the contribution of guided modes from the continuous spectrum.
• Fredholm Theory: The authors prove that the resulting system of integral equations is a Fredholm operator of index zero on the space $C_\alpha(\mathbb{R}) \oplus C_{\alpha+1/2}(\mathbb{R})$ for $0 < \alpha < 1/2$. This implies that the system is generically solvable, provided the null-space is trivial.
• Kernel Estimates: Detailed asymptotic estimates are derived for the boundary kernels $w^{[j]}(x_2, y_2)$ and their derivatives. These estimates characterize the behavior of the kernels both near the diagonal (singularities) and at infinity (decay rates), showing they satisfy specific asymptotic expansions involving $e^{ik_1|x|}$.
• Mode Projections: The paper demonstrates that the projection of the solution onto the guided wave-guide modes is determined entirely by the projections of the source terms $(\sigma, \tau)$ onto the mode functions. This allows for the approximate determination of scattering coefficients without solving the full integral system in certain regimes.
• Admissible Data: The framework accommodates various types of incoming data, including sums of wave-guide modes, point sources, and wave packets constructed from plane waves, provided they satisfy specific decay and asymptotic conditions.

Significance and Claims
The paper claims to provide a rigorous mathematical foundation for a numerical method to solve scattering problems in open wave-guide networks. By reformulating the problem as a transmission problem across an artificial boundary, the authors reduce the infinite domain problem to a system of integral equations on a line.

The authors explicitly state that:

1. The integral equations are Fredholm of index zero, making them suitable for numerical discretization and solution.
2. The method is well-adapted to numerical implementation, as noted in references to subsequent work [16, 17].
3. The formulation is symmetric and handles the interaction between guided modes and radiation fields naturally.
4. While this paper (Part I) establishes the existence of the integral equations and their Fredholm properties, the uniqueness of the solution (and thus the triviality of the null-space) relies on the specific outgoing radiation conditions defined in Part III of the series.
5. The approach avoids the direct application of the limiting absorption principle for the full network, instead using it locally to construct the fundamental solutions.

The paper concludes that this formulation offers a robust alternative to existing methods, particularly for problems involving wave-guide modes, and sets the stage for Parts II and III, which address the precise asymptotic behavior of the solutions and the proof of uniqueness via radiation conditions.