Abstract:This article presents novel numerical algorithms based on pseudodifferential operators for fast, direct, solution of the Helmholtz equation in 1D, 2D, and 3D inhomogeneous unbounded media. The proposed approach relies on an Operator Fourier Transform (OFT) representation of pseudodifferential operators ({\Psi}DO) which frame the problem of computing the inverse Helmholtz operator with a spatially-dependent wave speed in terms of two sequential applications of an inverse square root {\Psi}DO. The OFT representation of the action of the inverse square root {\Psi}DO, in turn, can be effected as a superposition of solutions of a pseudo-temporal initial-boundary-value problem for a paraxial equation. The OFT framework offers several advantages over traditional direct and iterative approaches for the solution of the Helmholtz equation. The operator integral transform is amenable to standard quadrature methods and the required pseudo-temporal paraxial equation solutions can be obtained using any suitable numerical method. A specialized quadrature is derived to evaluate the OFT efficiently and an alternating direction implicit method, used in conjunction with standard finite differences, is used to solve the requisite component paraxial equation problems. Numerical studies in 1D, 2D, and 3D are presented to confirm the expected OFT-based Helmholtz solver convergence rate. In addition, the efficiency and versatility of our proposed approach is demonstrated by tackling nontrivial wave propagation problems, including two-dimensional plane wave scattering from a geometrically complex inhomogeneity, three-dimensional scattering from turbulent channel flow and plane wave transmission through a spherically-symmetric gradient-index Luneburg lens. All computations, 3D problems which involve solving the Helmholtz equation with more than one billion complex unknowns, are performed in a single workstation.

On some Operator Filtering Strategies Based on Suitably Modified Green's Functions

An Improved Implementation of Multilevel Green's Function Interpolation Method for Electric Field Integral Equation

An OSRC Preconditioner for the EFIE

Two Sides Tangential Filtering Decomposition

Optimized Operator-Splitting Methods in Numerical Integration of Maxwell's Equations

Fourier-based schemes with modified Green operator for computing the electrical response of heterogeneous media with accurate local fields

Neural Green's Function Accelerated Iterative Methods for Solving Indefinite Boundary Value Problems

On Green's function embedding using sum-over-pole representations

Gradient Flows in Filtering and Fisher-Rao Geometry

Lattice Green's Functions for High Order Finite Difference Stencils

Extension and convergence analysis of Iterative Filtering to spherical data

Spectral Asymptotics of Elliptic Operators on Manifolds

A note on improvement by iteration for the approximate solutions of second kind Fredholm integral equations with Green's kernels

Enhancing Solutions for Complex PDEs: Introducing Complementary Convolution and Equivariant Attention in Fourier Neural Operators

Imposing different boundary conditions for thermal computational homogenization problems with FFT‐ and tensor‐train‐based Green's operator methods

Lattice Green's Functions for High-Order Finite Difference Stencils

EFIE Analysis of Low-Frequency Problems with Loop-Star Decomposition and Calderón Multiplicative Preconditioner

Operator approximation of the wave equation based on deep learning of Green's function

A novel direct Helmholtz solver in inhomogeneous media based on the operator Fourier transform functional calculus

Numerical solutions of Troesch's problem based on a faster iterative scheme with an application

Transverse Electric Scattering on Inhomogeneous Objects: Spectrum of Integral Operator and Preconditioning