🔢 mathematics

Embedded Trefftz DG method for steady Navier-Stokes flow. Part I: Oseen linearization

This paper develops and analyzes an embedded Trefftz-DG method for the Oseen problem, establishing stability and quasi-optimality through a novel local complement space construction and deriving a reduced velocity-only formulation to facilitate the analysis of the nonlinear Navier-Stokes equations in a subsequent part.

Original authors: Paul Stocker, Igor Voulis, Christoph Lehrenfeld, Philip L. Lederer

Published 2026-06-12

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Paul Stocker, Igor Voulis, Christoph Lehrenfeld, Philip L. Lederer

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 water flows through a complex pipe system. This is a classic problem in physics called the Navier-Stokes equations. However, these equations are incredibly difficult to solve because the water's speed affects how it pushes itself (a non-linear feedback loop).

To make this manageable, mathematicians often break the problem down into smaller, simpler steps. This paper focuses on the first, linear step of that process, known as the Oseen problem. Think of it as solving a "practice round" where the water's speed is frozen in place so we can figure out the basic flow patterns before tackling the full, chaotic reality.

The authors propose a new, smarter way to solve this using a method called Embedded Trefftz Discontinuous Galerkin (DG). Here is the concept broken down into everyday analogies:

1. The Old Way vs. The New Way

The Standard Approach (The "Brute Force" Method):
Imagine trying to describe a complex shape, like a cloud, by filling a grid with millions of tiny, identical Lego bricks. To get a smooth curve, you need so many bricks that the computer gets overwhelmed. This is what standard mathematical methods do: they use simple polynomial "bricks" (like flat or slightly curved surfaces) to approximate the flow. To get high accuracy, you need a massive number of them.

The New Approach (The "Smart Template" Method):
The authors use a technique called Trefftz. Instead of using generic Lego bricks, they use "smart templates" that are already shaped like the solution to the physics problem.

The Analogy: If you are trying to fit a puzzle piece into a hole that is shaped like a wave, a standard method tries to build a wave out of square blocks. The Trefftz method says, "Let's just use a piece that is already a wave."
The Benefit: Because these "smart templates" already obey the laws of physics (the differential equations), you need far fewer of them to get an accurate picture. It's like using a custom-molded cookie cutter instead of trying to carve a cookie out of a block of dough with a knife.

2. The "Embedded" Trick

There is a catch: Creating these perfect "wave-shaped" templates is very hard, especially when the water is flowing in different directions (variable coefficients). It's like trying to mold a cookie cutter for every single possible wind direction.

The authors' innovation is Embedded Trefftz.

The Analogy: Instead of trying to hand-craft the perfect cookie cutter, they start with a standard block of dough (a standard grid of polynomials). Then, they use a "sieve" or a "filter" to squeeze out only the parts of the dough that naturally form the right shape.
They don't explicitly build the perfect templates from scratch. Instead, they take a standard, easy-to-build grid and impose a rule: "You must satisfy the physics equation locally." This automatically selects the "smart" parts of the grid without the heavy lifting of constructing them manually.

3. The "Local-Global" Split

The paper introduces a clever way to organize the math, which they call Local-Global Splitting.

The Analogy: Imagine a large orchestra.
- The Global Part: This is the conductor and the main melody (the overall flow of the fluid). This is hard to coordinate because everyone needs to be in sync.
- The Local Part: These are the individual musicians practicing their own parts in their own rooms.
The authors' method separates the problem. They solve the "local" parts (the individual rooms) first. Because these rooms are independent, they can be solved very quickly and in parallel. Once the local parts are solved, they are "plugged in" to the global problem.
This reduces the size of the main problem significantly. Instead of solving one giant, messy puzzle, they solve many tiny, easy puzzles and then fit the results into a much smaller, cleaner main puzzle.

4. Why This Matters (The Results)

The paper proves two main things:

Stability: The method is mathematically sound. It won't crash or give nonsense answers, even when the flow gets tricky.
Efficiency: In their computer tests (using a standard flow pattern called Kovasznay flow), the new method achieved the same accuracy as the old method but used significantly fewer degrees of freedom (fewer "bricks").
- The Result: For the same level of accuracy, the new method was much faster. It's like getting a high-definition movie with a file size that fits on a USB stick, whereas the old method required a hard drive.

Summary

The authors have developed a mathematical "shortcut" for simulating fluid flow. By using a method that filters standard grids to find solutions that already obey the laws of physics, and by splitting the problem into easy local tasks and a small global task, they can solve complex flow problems much faster and with less computer power than traditional methods. This work lays the foundation for solving even more complex, real-world fluid problems (like turbulent air or blood flow) in a future paper.

Technical Summary: Embedded Trefftz DG Method for Steady Navier–Stokes Flow (Part I: Oseen Linearization)

Problem Statement
This paper addresses the numerical approximation of the steady Oseen problem, which serves as the linearization of the steady incompressible Navier–Stokes equations around a prescribed convection field $w$ . The governing equations seek a velocity $u$ and pressure $p$ satisfying:
$-\nu\Delta u + (w \cdot \nabla)u + \nabla p = f, \quad \text{div } u = 0$
in a bounded domain $\Omega \subset \mathbb{R}^d$ ( $d=2,3$ ), with homogeneous Dirichlet boundary conditions. The primary challenge addressed is the development of an embedded Trefftz discontinuous Galerkin (DG) method that maintains high approximation quality with reduced degrees of freedom (DoFs) while handling the variable coefficients introduced by the convection term $(w \cdot \nabla)u$ . Unlike the Stokes problem (where $w=0$ ), the Oseen operator involves variable coefficients, rendering standard constant-coefficient Trefftz constructions insufficient.

Methodology
The authors propose an embedded Trefftz-DG approach that avoids the explicit construction of local Trefftz bases. Instead, the method starts from a standard polynomial DG space $X$ and imposes the PDE locally through projection constraints.

Local Operator and Trefftz Space:
A local operator $K_{T,w}$ is defined on each element $T$ , mapping the solution space to a dual test space $Q(T)'$ . The Trefftz space $T_w$ is defined as the subspace of $X$ where the local residual vanishes in a weak sense against $Q(T)$ . Specifically, functions in $T_w$ satisfy the PDE only when tested against lower-order polynomials (degree $k-2$ for velocity, $k-1$ for pressure).
Local-Global Splitting and Complement Spaces:
A crucial theoretical step is the construction of a local complement space $L(T)$ such that $X(T) = T_w(T) \oplus L(T)$ .
- Velocity Component ( $L_u$ ): Constructed using bubble functions vanishing on a sub-ball, enriched with specific polynomial terms to control facet fluxes, and constrained to ensure zero mean flux on $d$ facets.
- Pressure Component ( $L_p$ ): Defined as zero-mean polynomials of degree $k-1$ .
- Invertibility: The paper proves that the local operator $K_{T,w}$ is boundedly invertible on $L(T)$ , provided the convection field $w$ satisfies a resolution condition (specifically, a smallness condition on a discrete norm $|w|_{h,d}$ relative to viscosity). This invertibility allows the local part of the solution to be computed independently on each element.
Reduced Formulation:
By eliminating the higher-order pressure components locally (which are tied to the velocity via the local operator), the authors derive a reduced formulation posed solely in terms of the velocity unknown. This formulation:
- Removes the saddle-point structure at the local level.
- Transfers coercivity from the Stokes setting to the reduced Oseen-Trefftz spaces.
- Retains a global coupling only for the piecewise-constant pressure component to enforce the discrete divergence constraint.
Stability Analysis:
The analysis establishes:
- Continuity of the bilinear forms.
- Coercivity of the reduced form on the Trefftz subspace.
- Inf-sup stability for the full coupled system in standard DG norms.
  These results rely on the abstract local-global framework and the specific construction of $L(T)$ , ensuring well-posedness and quasi-optimality.

Key Contributions
The paper extends the embedded Trefftz-DG methodology from the Stokes problem to the steady Oseen equations, providing the linear foundation for the subsequent nonlinear Navier–Stokes analysis (Part II). The specific contributions are:

Construction of Local Complement Spaces: Unlike the Stokes case, which relied on constant-coefficient Taylor-polynomial arguments, this work constructs $L(T)$ to handle the variable-coefficient local operator arising from the convection term.
Treatment of Inhomogeneous Right-Hand Sides: The method incorporates inhomogeneous terms directly into the local problem formulation rather than using a correction argument, fitting naturally into the abstract local-global splitting framework.
Reduced Formulation and Stability: The authors derive a reduced velocity-only formulation and prove reduced inf-sup stability for the full Oseen operator. This eliminates the saddle-point structure locally and allows for the transfer of coercivity under a suitable smallness condition on the convection field.

Results

Theoretical: The paper proves complete stability and quasi-optimality in standard DG norms. It establishes that the method is well-posed and that the error is bounded by the best approximation error in the discrete space, scaled by constants depending on the resolution of the convection field.
Numerical:
- Oseen–Kovasznay Test: Numerical experiments on the prescribed Oseen–Kovasznay flow demonstrate that the Trefftz-DG method achieves the expected asymptotic convergence rates ( $O(h^{k+1})$ for velocity, $O(h^k)$ for pressure).
- Efficiency: While the standard DG method yields slightly smaller error constants at fixed polynomial degree, the Trefftz-DG method achieves a given error level with significantly fewer degrees of freedom, particularly for higher polynomial orders.
- Computational Cost: Timing results indicate that the Trefftz-DG method is consistently faster than the standard DG method, with the advantage increasing on finer meshes and for higher orders.
- Limitations: The method shows sensitivity in the convection-dominated regime (low viscosity), where errors deteriorate earlier than in standard DG. This is attributed to the resolution and conditioning of the local embedded Trefftz construction.

Significance
The paper claims to provide the linear analytical groundwork necessary for the nonlinear Navier–Stokes theory developed in Part II. By successfully extending the embedded Trefftz approach to the Oseen problem with variable coefficients, the authors demonstrate that Trefftz techniques can be applied to inhomogeneous, variable-coefficient operators without explicit basis construction. The reduced formulation is highlighted as a crucial step for analyzing the nonlinear problem, as it decouples the local pressure from the global system, simplifying the stability analysis for iterative solvers. The work positions the Oseen problem as a bridge between classical DG discretizations and Trefftz-type model reduction, offering a pathway to high-order, low-DoF simulations for viscous incompressible flows.