A High Order Numerical Method for Scattering from Locally Perturbed Periodic Surfaces

TL;DR

This paper introduces an improved high order numerical method for scattering problems involving locally perturbed periodic surfaces, achieving super algebraic convergence and significantly enhancing computational efficiency over previous Bloch transform-based methods.

Contribution

The paper develops a novel numerical technique that leverages regularity results of the Bloch transform to improve convergence and efficiency in scattering problems with perturbed periodic surfaces.

Findings

Super algebraic convergence rate achieved.

Significant reduction in computation time and memory usage.

Potential applicability to 3D and electromagnetic scattering problems.

Abstract

In this paper, we will introduce a high order numerical method to solve the scattering problems with non-periodic incident fields and (locally perturbed) periodic surfaces. For the problems we are considering, the classical methods to treat quasi-periodic scattering problems no longer work, while a Bloch transform based numerical method was proposed in [LZ17b]. This numerical method, on one hand, is able to solve this kind of problems convergently; on the other hand, it takes up a lot of time and memory during the computation. The motivation of this paper is to improve this numerical method, from the regularity results of the Bloch transform of the total field, which have been studied in [Zha17]. As the set of the singularities of the total field is discrete in $\R$, and finite in one periodic cell, we are able to improve the numerical method by designing a proper integration contour with special conditions at the singularities. With a good choice of the transformation, we can prove that the new numerical method could possess a super algebraic convergence rate. \high{This new method improves the efficient significantly. At the end of this paper, several numerical results will be provided to show the fast convergence of the new method.} The method also provides a possibility to solve more complicated problems efficiently, e.g., three dimensional problems, or electromagnetic scattering problems.