Absorbing Boundary Conditions for Wave Propagation Simulations

145 concepts88 papers1 proposal

When simulating physical phenomena like seismic waves traveling through the Earth, light escaping a micro-laser, or acoustic waves propagating from a jet engine, the physical domain is effectively infinite. However, computer memory is finite. To run a simulation, we must truncate this infinite space into a finite computational box.

The moment we draw a hard boundary, we introduce a severe numerical artifact: spurious wave reflections. In nature, a wave would continue traveling outward forever. In a naive simulation, the edge acts like a mirror, bouncing energy back into the domain. As noted by Taeyoung Park in A method for removal of reflection artifact in computational fluid dynamic simulation of supersonic jet noise, these artificial reflections contaminate the primary data and can even create non-physical feedback loops that alter the source physics itself.

To solve this, researchers design Absorbing Boundary Conditions (ABCs) and Perfectly Matched Layers (PMLs). The goal is to make the artificial boundary completely transparent to outgoing waves, mimicking an open, unbounded universe within a finite grid.

The Physics of the Problem: Impedance Mismatch

Why do boundaries reflect waves? It comes down to impedance mismatch. When a wave encounters a sudden change in the medium's properties—or a sudden transition from a differential equation to a hard boundary condition (like a Dirichlet or periodic boundary)—the abrupt change forces a portion of the wave's energy to reflect backward.

As explained by Georg Maierhofer et al. in Computing nonlinear Schrödinger equations with Hermite functions beyond harmonic traps, standard periodic boundaries fail to capture natural dispersive behavior, trapping energy inside the box and causing non-physical interference patterns. To prevent this, the boundary must be mathematically engineered to absorb energy without creating a sharp transition.

Mechanism 1: One-Way Wave Equations

One of the earliest insights for eliminating reflections is to change the governing physics at the boundary so that waves can only travel in one direction: outward.

A standard wave equation allows for bidirectional propagation (left and right, or in and out). By mathematically factorizing the wave operator into two directional components, we can isolate the outgoing wave component. This is known as a paraxial approximation.

M. Yu. Trofimov in A nonstationary form of the range refraction parabolic equation and its application as an artificial boundary condition for the wave equation in a waveguide demonstrates how factorizing the wave operator allows for the construction of artificial boundaries that selectively absorb outgoing waves. Similarly, Yunyin Dai in Finite-Difference Solution of the Helmholtz Equation Based on Two Domain Decomposition Algorithms utilizes paraxial absorbing boundary conditions to simulate infinite media in complex seismic environments, allowing waves to exit the computational domain seamlessly.

Mechanism 2: Perfectly Matched Layers (PML)

Instead of modifying the boundary condition at a single sharp interface, what if we surround the computational domain with an artificial "buffer zone" that absorbs waves gradually? This is the core idea behind the Perfectly Matched Layer (PML).

A PML is an artificial layer designed with two critical properties:

  1. Perfect Impedance Matching: At the interface between the real domain and the PML, the wave impedance matches perfectly regardless of the angle of incidence. The wave enters the layer with zero reflection.
  2. Exponential Attenuation: Inside the PML, the equations are modified (often via complex coordinate stretching) so that the wave's amplitude decays exponentially as it moves deeper into the layer.

In Seismo-VLAB: An Open-Source Software for Soil–Structure Interaction Analyses, Danilo S. Kusanovic et al. demonstrate that combining PMLs with domain reduction methods is essential for accurately simulating soil-structure interactions without non-physical reflections. Tianpeng Jiang further applies this in Perfectly-matched-layer method for optical modes in dielectric cavities to truncate infinite space when calculating leaky optical resonances, avoiding the physical paradox of infinitely growing wave solutions in open systems.

Mechanism 3: Viscoelastic Boundaries

In structural engineering and geomechanics, waves propagate through solid media. Here, simply absorbing energy is not enough; the boundary must also maintain the structural and static integrity of the model.

If you use a purely viscous boundary (just dampers), it will absorb wave energy but lack a restoring force, causing the entire model to drift or slide over time under continuous loads. To fix this, researchers use viscoelastic boundaries, which pair dampers with springs.

As evaluated by Dan Ye et al. in Semi-infinite Domain Artificial Boundary Conditions in Dynamic Analysis, the dampers dissipate the incident wave energy to prevent artificial reflections, while the springs provide the necessary static stiffness to prevent model drift. This combination ensures both dynamic accuracy and long-term numerical stability in finite element analysis.

The Discretization Trap: Numerical Reflections

An elegant mathematical formula for an ABC does not automatically guarantee a reflection-free simulation. When we discretize continuous equations onto a grid (using finite differences or finite elements), the grid itself introduces a new source of reflection: numerical impedance mismatch.

If the artificial boundary condition is not mathematically aligned with the internal semi-discretization scheme, waves will still reflect off the boundary due to discrete approximation errors.

To eliminate this, Songsong Ji et al. show in Accurate artificial boundary conditions for semi-discretized one-dimensional peridynamics that designing ABCs to be recursively consistent with the internal discretization scheme ensures numerical exactness, allowing waves to pass through seamlessly. Similarly, Jerry Zhijian Yang et al. in Stability and convergence analysis of high-order numerical schemes with DtN-type absorbing boundary conditions for nonlocal wave equations utilize high-order Dirichlet-to-Neumann (DtN) conditions to match boundary values to internal nonlocal fluxes, preserving spatial accuracy without boundary artifacts.

Alternative Paradigms: Bypassing Truncation Entirely

Rather than trying to fix the boundaries of a truncated box, some modern approaches reformulate the problem to avoid hard truncation altogether.

One method is to use Boundary Integral Equations (BIEs), which mathematically project the behavior of the infinite exterior domain directly onto the boundary of the region of interest. In Learning Only On Boundaries: a Physics-Informed Neural operator for Solving Parametric Partial Differential Equations in Complex Geometries, Zhiwei Fang et al. leverage boundary integrals within a neural network framework, naturally satisfying radiation conditions at infinity without volume sampling.

Furthermore, Shunyuan Mao et al. demonstrate in Neural Networks as Surrogate Solvers for Time-Dependent Accretion Disk Dynamics that embedding physical laws directly into a neural network's loss function can naturally bypass the spurious reflection problem that frequently plagues traditional grid-based solvers.

Key Foundational Work

Complex Riemannian metric and absorbing boundary conditions (2001) — contributed 4 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 86 papers
  2. Version 1 · 2026-10-04 · 133 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_artificial_boundary_reflection_3d6aaaf9,
  title = {artificial boundary reflection},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/artificial-boundary-reflection/rdr_3d6aaaf9bc22482e855a26561d3d34a6},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 2, frozen text):

@misc{lacuna_artificial_boundary_reflection_3d6aaaf9_v2,
  title = {artificial boundary reflection},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/artificial-boundary-reflection/rdr_3d6aaaf9bc22482e855a26561d3d34a6/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake