🔢 mathematics

Weak and strong averaging principle for 2D Boussinesq equations with non-Lipschitz Poisson jump noise

This paper establishes the well-posedness, ergodicity, and both weak and strong averaging principles for 2D Boussinesq equations driven by non-Lipschitz Poisson jump noise, substantiated by a specific case study and numerical simulations.

Original authors: Yangyang Shi, Dong Su, Hui Liu

Published 2026-02-06

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yangyang Shi, Dong Su, Hui Liu

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 watching a complex dance between two partners: a slow, heavy dancer (representing the swirling motion of a fluid, called vorticity) and a fast, jittery dancer (representing the temperature of that fluid).

In the real world, these two are constantly influencing each other. The slow dancer moves the fast one, and the fast one pushes back. But the fast one is also being shaken by a chaotic, unpredictable force—like a crowd of people randomly bumping into them (this is the "Poisson jump noise").

The paper by Shi, Su, and Liu tackles a very specific mathematical challenge: Can we ignore the fast dancer's chaotic details and just predict the slow dancer's path by looking at the "average" effect of the fast one?

Here is a breakdown of their work using simple analogies:

1. The Setup: A Two-Speed System

The authors are studying the 2D Boussinesq equations. Think of this as the rulebook for how heat and fluid move together (like hot air rising or ocean currents).

The Slow Variable ( $j^\epsilon$ ): This is the fluid's spin or "vorticity." It changes slowly.
The Fast Variable ( $\theta^\epsilon$ ): This is the temperature. It changes very quickly.
The Noise: The system isn't smooth; it's being hit by random "jumps" (like sudden gusts of wind or random thermal spikes). Crucially, the math describing these jumps is "non-Lipschitz." In plain English, this means the rules for how the jumps behave are a bit "rough" or "jagged," making them much harder to predict than standard smooth curves.

2. The Problem: Too Much Detail to Handle

If you try to simulate this system on a computer, you have to calculate every single tiny, rapid movement of the temperature to know where the fluid spin will be next. This is computationally exhausting and mathematically messy, especially with the "rough" noise.

The Averaging Principle is a shortcut. It suggests that if the temperature changes fast enough, the slow fluid spin doesn't care about the exact moment-to-moment temperature. It only cares about the average temperature over time.

3. What the Authors Did

The paper proves that this shortcut works, even with the "rough" noise and the complex interaction between the two dancers.

Step 1: Proving the Dancers Exist and Stay Stable.
First, they had to prove that the system actually makes sense mathematically. They showed that even with the rough noise, the fluid spin and temperature won't explode to infinity or behave wildly. They proved the "fast" temperature dancer eventually settles into a predictable statistical pattern (called ergodicity). Imagine the fast dancer spinning so fast that, over a long time, they cover every spot on the floor equally.
Step 2: The "Freezing" Trick.
To find the average, they used a technique called "freezing." They pretended the slow dancer stood still for a moment and watched how the fast dancer behaved. They proved that the fast dancer's behavior stabilizes quickly, allowing them to calculate a single "average force" that the fast dancer exerts on the slow one.
Step 3: The Convergence (The Main Result).
They proved that as the time-scale difference gets larger (meaning the temperature gets infinitely faster compared to the fluid spin), the actual path of the slow dancer gets closer and closer to the path predicted by the Averaged Equation.
- Weak Convergence: The probability of the slow dancer being far off from the average path goes to zero.
- Strong Convergence: The average distance between the real path and the predicted path goes to zero.

4. The "Roughness" Challenge

Most previous math papers assumed the noise was smooth (like a gentle breeze). This paper is special because it handles non-Lipschitz noise.

Analogy: Imagine the noise isn't a gentle breeze, but a series of random, sharp pinpricks. The math for pinpricks is "rougher" and harder to handle. The authors developed new tools to show that even with these sharp, jagged random hits, the averaging shortcut still works.

5. The Proof in Action (Simulation)

To make sure their math wasn't just theory, they created a specific example with a "jagged" noise function (using fractional powers like $|j|^{2/3}$ ).

They ran computer simulations with different speeds for the fast dancer.
The Result: As they made the fast dancer faster and faster, the error between the real simulation and their simplified "averaged" prediction got smaller and smaller, vanishing almost completely.

Summary

In simple terms, this paper says: "Even if a fluid system is being hit by random, jagged, unpredictable shocks, and the temperature is changing wildly fast, you can safely ignore the tiny, chaotic details of the temperature. You can replace the complex, fast-moving temperature with a simple 'average' value, and your prediction for the fluid's movement will be incredibly accurate."

This allows scientists and engineers to use much simpler, faster models to predict complex fluid behaviors without losing accuracy.

Technical Summary: Weak and Strong Averaging Principle for 2D Boussinesq Equations with Non-Lipschitz Poisson Jump Noise

Problem Statement
This paper investigates the averaging principle for two-dimensional (2D) Boussinesq equations defined on a torus $T^2$ , subject to non-Lipschitz Poisson jump noise. The system models fluid flow and heat transport where the velocity field $u$ and temperature field $\theta$ interact. The study focuses on a two-scale stochastic system where the temperature variable evolves on a fast time scale relative to the vorticity variable $j$ (defined as the curl of the velocity). The governing equations are driven by compensated Poisson random measures, and the coefficients satisfy non-Lipschitz conditions, specifically characterized by a concave, continuous, non-decreasing function $\kappa(\cdot)$ rather than standard Lipschitz bounds. The primary objective is to rigorously prove that as the time-scale separation parameter $\varepsilon \to 0$ , the slow variable (vorticity) converges to the solution of an averaged equation.

Methodology
The authors employ a multi-step analytical framework combining stochastic analysis, functional analysis, and ergodic theory:

Well-posedness and Regularity: The paper first establishes the existence and uniqueness of solutions for the original two-scale system under non-Lipschitz conditions. Using Itô's formula for Lévy noise and the Burkholder-Davis-Gundy (B-D-G) inequality, the authors derive uniform regularity estimates and tightness properties for the vorticity variable $j^\varepsilon$ .
Ergodicity of the Frozen Equation: To construct the averaged equation, the authors analyze the "frozen" equation for the fast temperature variable $\theta$ , where the slow variable $j$ is treated as a fixed parameter. They prove the existence of a unique ergodic invariant measure $\pi^u$ for this frozen system and establish exponential convergence to this equilibrium.
Construction of the Averaged Equation: The averaged coefficients ( $\bar{f}, \bar{\sigma}_1, g$ ) are defined as integrals of the original coefficients with respect to the invariant measure $\pi^u$ . The paper demonstrates that these averaged coefficients retain the necessary non-Lipschitz continuity properties to ensure the well-posedness of the limiting averaged equation.
Convergence Proofs:
- Weak Convergence: Using the tightness of the family of distributions and the Skorohod representation theorem, the authors prove that the vorticity variable converges in probability to the solution of the averaged equation.
- Strong Convergence: Under the specific condition that the noise term in the slow equation ( $\sigma_1$ ) does not depend on the fast variable $\theta$ , the authors utilize Khasminskii's time-discretization method. They construct an auxiliary process and employ Bihari's inequality to prove strong convergence in the $2p$ -th mean.

Key Contributions and Results

Extension to Non-Lipschitz Noise: The paper generalizes the averaging principle to stochastic Boussinesq equations driven by Poisson jump noise with non-Lipschitz coefficients, a setting that presents significant challenges compared to standard Lipschitz or Gaussian noise frameworks.
Rigorous Convergence Theorems:
- Theorem 5.2 (Weak Convergence): It is proven that for any $T > 0$ and $\eta > 0$ , $\lim_{\varepsilon \to 0} P(\sup_{t \in [0,T]} |j^\varepsilon - \bar{j}|_H > \eta) = 0$ .
- Theorem 5.3 (Strong Convergence): Under the assumption that the slow noise is independent of the fast variable, strong convergence is established: $\lim_{\varepsilon \to 0} E[\sup_{t \in [0,T]} |j^\varepsilon_t - \bar{j}_t|^{2p}_H] = 0$ .
Ergodicity Analysis: The study provides detailed estimates on the ergodicity of the frozen temperature equation, including bounds on the invariant measure and convergence rates, which are crucial for defining the averaged drift and diffusion terms.
Numerical Validation: A specific example involving non-Lipschitz power-law terms ( $|j|^{2/3}$ , $|\theta|^{1/3}$ ) is constructed. Numerical simulations with 100 samples confirm the theoretical predictions, showing that the mean squared error decays as $\varepsilon$ decreases, consistent with the derived convergence rates.

Significance
The paper addresses a gap in the literature regarding multi-scale stochastic fluid dynamics, specifically where fast variables satisfy reaction-diffusion equations with convective terms and non-Lipschitz noise. The authors note that existing results often require abstract reaction-diffusion structures that differ from the Boussinesq model. By handling the convective terms and non-Lipschitz conditions simultaneously, this work provides a robust theoretical foundation for approximating complex multi-scale fluid-thermal systems. The results are significant for understanding the statistical behavior of such systems over long time scales and for developing reduced-order models that capture the essential dynamics of the slow variables while averaging out the fast fluctuations. The numerical examples substantiate the theoretical findings, demonstrating the practical applicability of the derived averaging principles.