🔢 mathematics

A HDG method with transmission variables for time-harmonic wave propagation problems with constant coefficients

This paper extends the CHDG method, a hybridizable discontinuous Galerkin variant using transmission variables that accelerates iterative convergence, to a general framework for time-harmonic wave propagation with constant coefficients, demonstrating its effectiveness and well-posedness through applications to the linearized Euler equations with uniform subsonic mean flow.

Original authors: Simone Pescuma, Gwénaël Gabard, Théophile Chaumont-Frelet, Axel Modave

Published 2026-06-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Simone Pescuma, Gwénaël Gabard, Théophile Chaumont-Frelet, Axel Modave

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to predict how sound waves, wind, or ripples in water move through a complex environment. In the world of physics and engineering, this is like trying to track a massive, chaotic crowd of people moving through a maze. Doing this with computers is incredibly hard because the waves wiggle and vibrate so fast that the math gets messy, and the computers take forever to find the answer.

This paper introduces a new, smarter way to solve these "wave problems" using a method called CHDG. Here is a simple breakdown of what they did and why it matters, using everyday analogies.

The Problem: The "Slow Crowd"

Think of a standard computer simulation of waves like a group of people trying to cross a room by passing a message from person to person.

The Old Way (Standard Methods): The computer breaks the room into tiny tiles (like a mosaic). To solve the problem, every tile has to talk to its neighbors, and the computer has to solve a giant, tangled web of equations all at once. It's like trying to get a message to the other side of the room by having everyone shout at once; it gets noisy, slow, and the computer gets stuck in a loop trying to figure out the answer.
The Bottleneck: The math behind these wave problems is "unstable" in a way that makes iterative solvers (computers that guess and check) very slow to converge. They take thousands of steps to get close to the right answer.

The Solution: The "Transmission Variable"

The authors developed a new strategy called CHDG (Characteristic Hybridizable Discontinuous Galerkin). Instead of treating the whole room as one giant puzzle, they changed how the tiles talk to each other.

The Analogy: The Relay Race vs. The Handshake

Standard Method (The Handshake): In the old way, when two tiles meet, they try to agree on the exact physical state (like pressure or speed) at the boundary. It's like two neighbors arguing over the exact temperature of the wall between their houses. It's a lot of back-and-forth.
The New Method (The Relay Race): The CHDG method introduces a special "Transmission Variable." Think of this as a messenger running along the walls between the tiles.
- Instead of arguing about the state of the wall, the tiles only pass a specific "message" to their neighbors.
- This message is based on the physics of the wave: "I am sending a wave out to you, and I am receiving a wave from you."
- By focusing only on these incoming and outgoing messages, the computer can solve the problem for each tile independently and then just swap the messages.

Why It's Faster: The "Strict Contraction"

The paper proves mathematically that this new way of passing messages has a special property called a "strict contraction."

The Metaphor: Imagine you are trying to find a specific spot in a dark room by taking steps. In the old method, you might take a step forward, then a step back, then a step forward again, wandering around for a long time before finding the spot.
The New Method: With the CHDG method, every time you take a step (an iteration), you are guaranteed to get strictly closer to the target. You never wander back. The math proves that the "distance" to the correct answer shrinks with every single step. This means the computer finds the answer in far fewer steps than before.

The Real-World Test: Sound and Swirls

To prove this works, the authors tested it on a specific, tricky problem: Aeroacoustics (sound moving through wind).

The Challenge: In a room with wind blowing through it, there are two types of "waves":
1. Sound Waves: Like normal noise, moving in all directions.
2. Vorticity Waves: Like tiny swirls or eddies in the wind. These are tricky because they only move with the wind, not against it.
The Result: The new method handled both types of waves perfectly. It didn't matter if the wind was blowing left or right; the "messenger" system adapted automatically.
The Comparison: They compared their new method against the old standard methods using different computer algorithms (like Fixed-Point, CGNR, and GMRES).
- The Winner: The new CHDG method was consistently faster. The "Fixed-Point" method (the simplest guess-and-check approach) worked surprisingly well and converged almost as fast as the most complex algorithms, but with much less memory usage.

Summary

In short, this paper says: "We found a way to break down complex wave problems into smaller, independent pieces that talk to each other via specific 'incoming' and 'outgoing' messages. This structure guarantees that the computer gets closer to the answer with every single step it takes, making the simulation much faster and more efficient than the old ways of doing it."

They didn't just prove it on paper; they ran simulations showing it works for both sound waves and wind swirls, proving it's a robust tool for engineers dealing with complex wave physics.

Technical Summary: A HDG Method with Transmission Variables for Time-Harmonic Wave Propagation

Problem Statement
Iterative finite element solvers for time-harmonic wave propagation problems are notoriously slow to converge due to the intrinsic properties of the underlying linear systems. While Discontinuous Galerkin (DG) methods are well-suited for high-fidelity solutions in complex geometries and high-frequency regimes, the resulting sparse unstructured linear systems pose significant challenges for direct solvers on parallel architectures. Although iterative solvers offer memory efficiency, their convergence rates are often insufficient. Existing Hybridizable Discontinuous Galerkin (HDG) methods improve system conditioning by eliminating physical unknowns within elements, leaving a smaller system defined on the mesh skeleton. However, standard HDG formulations, which utilize numerical traces as hybrid variables, do not always yield the fastest convergence for iterative schemes in wave propagation contexts.

Methodology
The authors propose a variant of the HDG method, termed CHDG (Characteristic-based HDG), which replaces the standard numerical trace with transmission variables derived from the characteristic analysis of the governing hyperbolic systems.

Mathematical Framework: The method targets linear, symmetric, hyperbolic systems with constant coefficients in the frequency domain. The governing equations are analyzed via characteristic decomposition, splitting the solution into outgoing ( $G^+$ ) and incoming ( $G^-$ ) transmission variables based on the eigenvalues of the flux matrix.
DG Formulation: A DG scheme is constructed using upwind fluxes expressed explicitly in terms of these transmission variables. The numerical flux at an interface utilizes the outgoing variable from the current element and the incoming variable from the neighboring element.
Hybridization: The core innovation lies in the hybridization strategy. Instead of eliminating physical variables via a numerical trace, the method introduces the incoming transmission variables ( $G^-$ $G^{-}$ ) as the global unknowns on the mesh skeleton.
- Physical variables within each element are eliminated by solving local problems where $G^-$ is treated as a boundary datum.
- This results in a global hybridized system of the form $(I - \Pi S)G^- = b$ , where $S$ is a scattering operator (mapping incoming to outgoing variables) and $\Pi$ is an exchange operator (handling interface coupling and boundary conditions).
Theoretical Properties: The authors prove that for passive boundary conditions, the operator $\Pi S$ is a strict contraction. Consequently, the hybridized system is well-posed and amenable to fixed-point iteration.

Application to Aeroacoustics
The framework is specifically applied to the time-harmonic linearized Euler equations (LEE) modeling wave propagation in a uniform subsonic mean flow. This application introduces specific complexities:

The system supports both sound waves and vorticity waves.
The number of incoming and outgoing transmission variables on an element face is not constant; it depends on the orientation of the face relative to the mean flow velocity ( $u_0 \cdot n$ ).
The proposed general framework naturally handles this variability in the number of boundary conditions and transmission variables across the domain.

Numerical Results
The method is validated through two-dimensional numerical benchmarks involving sound and vorticity waves in uniform mean flows, comparing the convergence of three iterative schemes (Fixed Point, CGNR, and GMRES) applied to both standard DG and CHDG systems:

Plane Wave Propagation: Tests with varying propagation directions relative to the mean flow and different boundary conditions (impedance vs. pressure).
Point Source: A singular source problem in a circular domain with varying Mach numbers.
Vorticity Waves: Propagation in a rigid waveguide with higher-order modes.

Key Findings:

The Fixed Point iteration applied to the CHDG system consistently converges, aligning with the theoretical contraction proof.
Iterative schemes (CGNR and GMRES) applied to the CHDG system converge significantly faster (fewer iterations) than when applied to standard DG systems.
In the CHDG context, the Fixed Point method performs nearly as well as GMRES and consistently outperforms CGNR.
The presence of vorticity waves and varying flow orientations does not hinder the convergence properties of the CHDG approach.

Significance and Claims
The paper claims to extend the CHDG methodology, previously established for Helmholtz problems, to a general framework covering a broad class of wave propagation problems defined by symmetric hyperbolic systems with constant coefficients.

Unified Handling: The framework unifies the treatment of diverse wave types (acoustic and vorticity) within a single mathematical structure.
Robustness: It effectively manages the complexity of variable numbers of transmission variables caused by mean flow orientation, a feature not naturally handled by previous specific implementations.
Efficiency: The reformulation significantly accelerates the convergence of iterative solvers compared to standard DG and HDG approaches, making it a promising candidate for large-scale, high-frequency wave propagation simulations.

The authors note that while the current results are based on MATLAB scripts, preliminary work on electromagnetic waves suggests similar benefits, and future work aims to extend this framework to elastic waves and non-homogeneous media.