Boundary integral equations for calculating complex eigenvalues of transmission problems

TL;DR

This paper introduces new boundary integral equations for transmission problems that effectively distinguish true resonance frequencies from fictitious ones, improving the accuracy of eigenvalue computations in waveguide and scattering problems.

Contribution

The paper proposes novel boundary integral equations that can differentiate true eigenvalues from fictitious ones in transmission problems, enhancing the reliability of resonance frequency calculations.

Findings

Proposed BIEs successfully separate true and fictitious eigenvalues.

Numerical verification in 2D and 3D problems confirms effectiveness.

True eigenvalues relate significantly to solution behavior.

Abstract

Resonance frequencies are complex eigenvalues at which the homogeneous transmission problems have non-trivial solutions. These frequencies are of interest because they affect the behavior of the solutions even when the frequency is real. The resonance frequencies are related to problems for infinite domains which can be solved efficiently with the Boundary Integral Equation Method (BIEM). We thus consider a numerical method of determining resonance frequencies with fast BIEM and the Sakurai-Sugiura projection Method (SSM). However, BIEM may have fictitious eigenvalues even when one uses M\"uller or PMCHWT formulations which are known to be resonance free when the frequency is real valued. In this paper, we propose new BIEs for transmission problems with which one can distinguish true and fictitious eigenvalues easily. Specifically, we consider waveguide problems for the Helmholtz equation in 2D and standard scattering problems for Maxwell's equations in 3D. We verify numerically that the proposed BIEs can separate the fictitious eigenvalues from the true ones in these problems. We show that the obtained true complex eigenvalues are related to the behavior of the solution significantly. We also show that the fictitious eigenvalues may affect the accuracy of BIE solutions in standard boundary value problems even when the frequency is real.