The Semigeostrophic-Euler Limit: Lifespan Lower Bounds and $O(\varepsilon)$ Velocity Stability

This paper establishes strong $O(\varepsilon)$ stability estimates for the velocity and physical densities of the two-dimensional semigeostrophic system relative to the incompressible Euler equations on a flat torus, while also proving a lifespan lower bound of $T_*(\varepsilon) \gtrsim \varepsilon^{-1}\log\log(1/\varepsilon)$ that improves upon the standard hyperbolic scale.

The Gist

Imagine the Earth's atmosphere and oceans as a giant, swirling dance floor. Scientists use complex math to predict how the air and water move. For large-scale weather patterns, they often use a simplified model called the Semigeostrophic (SG) system. It's like a "cheat code" that makes the math easier by assuming the spinning of the Earth (the Coriolis force) is the main boss, overpowering the inertia of the moving air.

However, this cheat code isn't perfect. It's an approximation. The "real" physics is described by the Euler equations, which are much more complicated and accurate but also much harder to solve.

This paper, written by Victor Armengou, asks a very practical question: "How long can we trust the 'cheat code' (SG) before it starts to drift too far away from the 'real physics' (Euler)?"

Here is the breakdown of the paper's findings using simple analogies:

1. The Setup: Two Runners on a Track

Imagine two runners starting a race from the exact same spot:

• Runner A (Euler): Runs the "real" path. They are fast, accurate, but their path is determined by a very complex, non-linear rule (like a maze that changes as they run).
• Runner B (SG): Runs the "approximate" path. They follow a slightly simpler rule. At the very beginning, they are almost perfectly side-by-side with Runner A.

The paper studies what happens when Runner B tries to stay close to Runner A. The author introduces a small parameter, $\epsilon$ (epsilon), which represents how "small" the initial difference is. Think of $\epsilon$ as the size of a tiny pebble in the shoe of Runner B. The smaller the pebble, the closer they start.

2. The Big Discovery: The "Double-Exponential" Lifespan

Usually, when you use an approximation in physics, it works well for a short time, and then the errors pile up quickly. You might expect the two runners to drift apart after a time proportional to $1/\epsilon$.

The paper's first major finding is a surprise:
Runner B stays close to Runner A for much longer than expected.

• Standard Expectation: They drift apart after time $T \approx 1/\epsilon$.
• Paper's Result: They stay close until $T \approx \frac{1}{\epsilon} \times \log(\log(1/\epsilon))$.

The Analogy:
Imagine you are walking on a tightrope. Usually, if you wobble a tiny bit, you might fall off in a few seconds. This paper proves that because of the specific way the "tightrope" (the math of the atmosphere) is built, you can wobble for a very long time—specifically, a time that is longer by a "log-log" factor. It's like finding out that the tightrope has a hidden safety net that keeps you balanced for an extra hour, even though you thought you'd fall in a minute.

3. The "Speedometer" Check (Velocity Stability)

The authors didn't just say "they stay close." They proved that the speed of the approximate runner (SG) is almost identical to the real runner (Euler).

• If the initial error is small ($\epsilon$), the difference in their speeds stays small ($O(\epsilon)$) for that entire extended time.
• Analogy: If Runner A is running at 10 mph, and Runner B is running at $10 + 0.001$ mph, the paper proves that Runner B won't suddenly speed up to 20 mph or slow down to 2 mph. They stay within that tiny margin of error for a surprisingly long time.

4. The "Crowd Density" Check (Wasserstein Distance)

In fluid dynamics, we care not just about speed, but about where the "stuff" (air or water) is located.

• The paper proves that if you look at the density of the crowd (where the air molecules are), the SG model and the Euler model are also very close.
• The Analogy: Imagine two groups of people walking through a city. One group follows the complex rules of the city (Euler), and the other follows a simplified map (SG). The paper proves that even though the simplified map is an approximation, the two groups of people will end up in almost the exact same neighborhoods, and the "clumps" of people will look almost identical.

5. How Did They Do It? (The Secret Sauce)

The authors used a clever mix of tools:

• The "Bootstrap" Method: They assumed the approximation holds for a while, and then proved that if it holds for a little while, it must hold for a little longer. It's like pushing a swing; if you push it just right, it keeps going.
• Optimal Transport: They used a branch of math that deals with the most efficient way to move things from point A to point B. They treated the air density like a pile of sand that needs to be moved, and used these "moving sand" rules to track how the two models diverge.
• The "Determinant" Trick: The SG model has a tricky math term (the Monge-Ampère equation) that makes it non-linear. The authors showed that when the errors are small, this tricky term behaves almost like a simple linear equation, which is much easier to control.

Summary

In plain English, this paper is a reassurance for weather forecasters and oceanographers.

It says: "You can use the simplified Semigeostrophic model to predict large-scale weather and ocean currents. Even though it's an approximation, it won't go wrong quickly. In fact, it stays accurate for a surprisingly long time, and the speed and location of the air/water it predicts will be very close to the 'perfect' physics model."

It turns a "good enough for now" approximation into a "reliable for a long time" tool, giving scientists more confidence in their long-term climate and weather models.

Technical Summary

1. Problem Statement and Context

The Model:
The paper investigates the Semigeostrophic (SG) system, a classical model in geophysical fluid dynamics describing large-scale rotating flows in the atmosphere and ocean. The system is derived from the incompressible Euler equations under the Boussinesq and hydrostatic approximations, where the Coriolis force dominates inertial terms.

The Mathematical Challenge:
In its "dual" formulation (using geostrophic variables), the SG system presents a unique challenge: the velocity field is determined not by a linear Biot-Savart law (as in standard Euler), but by a fully nonlinear Monge-Ampère equation.

• Dual Variables: The unknowns are a density $\rho$ (with unit mean) and a convex potential $\Psi$. The velocity is $u = (\nabla \Psi - x)^\perp$.
• Coupling: The density and potential are linked by $\det(I + D^2\psi) = \rho$, where $\psi = \Psi - |x|^2/2$.

The Limit Question:
The paper focuses on the small-amplitude regime, where the density is a small perturbation of a constant state ($\rho = 1 + \varepsilon \omega$) and the potential is a small perturbation of the quadratic map. Formally, as the small parameter $\varepsilon \to 0$, the SG system converges to the 2D Incompressible Euler equations.
The paper aims to quantify this convergence rigorously by answering three specific questions:

1. Lifespan: How long does the perturbative regime (where SG is close to Euler) persist in physical time?
2. Velocity Stability: Can we establish a strong $L^2$ estimate for the difference between SG and Euler velocities?
3. Density Comparison: Can we compare the physical densities of the two systems directly using the Wasserstein distance, even without assuming Lipschitz regularity for the SG velocity?

2. Methodology and Framework

The author works on the flat 2D torus $\mathbb{T}^2$ using the dual (geostrophic) formulation within Loeper's framework.

Key Analytical Tools:

• Bootstrap Regime: The analysis assumes a "bootstrap" condition where the Monge-Ampère term remains a small perturbation of the Laplacian: $\varepsilon \|\nabla \rho_\varepsilon\|_{L^\infty} \leq 1/4$.
• Endpoint Calderón-Zygmund Estimates: Used to control the Hessian of the potential ($D^2\psi$) in $L^\infty$ based on the density regularity. This is crucial for handling the logarithmic loss of derivatives typical in 2D Euler.
• Wente-Type Inequalities: Used to map the quadratic Monge-Ampère term ($\det D^2\psi$) into the $H^{-1}$ space, allowing it to be treated as a controlled error term in energy estimates.
• Flow-Based Stability (Loeper's Estimate): Instead of comparing velocities directly (which loses derivatives), the author compares the Lagrangian flows ($X_{SG}$ and $X_{Euler}$). The difference in flows is linked to the difference in densities via optimal transport stability estimates.
• Representation Principles:
  • Deterministic: For the smooth Euler solution (Lipschitz velocity).
  • Probabilistic (Superposition): For the SG solution, utilizing the fact that the continuity equation holds in a weak sense even if the velocity is not Lipschitz.

3. Key Contributions and Results

A. Lifespan Lower Bound (Log-Log Improvement)

Result: The paper proves that the perturbative regime persists for a physical time $T^*(\varepsilon)$ satisfying:
$$T^*(\varepsilon) \gtrsim \varepsilon^{-1} \log \log (1/\varepsilon)$$
Significance:

• Standard hyperbolic scaling suggests a lifespan of $O(\varepsilon^{-1})$.
• By combining transport estimates with endpoint Calderón-Zygmund bounds, the author derives a Riccati-type inequality with logarithmic nonlinearity ($\dot{y} \lesssim y \log y$) rather than quadratic nonlinearity.
• This yields a double-exponential growth in slow time, translating to a log-log improvement in physical time. This extends the validity of the SG-Euler approximation significantly beyond the standard scale.

B. Strong $O(\varepsilon)$ Velocity Stability

Result: On the bootstrap window, the difference between the SG velocity $u_\varepsilon$ and the Euler velocity $\bar{u}$ is bounded in $L^2$:
$$\| u_\varepsilon(t) - \bar{u}(t) \|_{L^2(\mathbb{T}^2)} \leq C_t \varepsilon$$
Methodology:

• The proof avoids direct energy estimates on the velocity difference (which would fail due to derivative loss).
• Instead, it establishes a differential inequality for the flow gap $\|X_\varepsilon - \bar{X}\|_{L^2}$.
• The forcing term in this inequality is $O(\varepsilon)$, arising from the determinantal correction $\varepsilon \det D^2\psi_\varepsilon$.
• Using Grönwall's inequality and the uniform $L^2$ bound on the Hessian (derived from the bootstrap), the flow gap is shown to be $O(\varepsilon)$, which implies the velocity bound via Loeper's stability estimate.

C. Wasserstein Comparison of Physical Densities

Result: A general comparison theorem is proven for the physical densities $m_\varepsilon = 1 + \varepsilon \rho_\varepsilon$ and $\bar{m}_\varepsilon = 1 + \varepsilon \bar{\rho}$:
$$W_2^2(m_\varepsilon(t), \bar{m}_\varepsilon(t)) \leq e^{A(t)} W_2^2(m_\varepsilon(0), \bar{m}_\varepsilon(0)) + \int_0^t e^{A(t)-A(s)} \| u_\varepsilon(s) - \bar{u}(s) \|_{L^2(m_\varepsilon)}^2 ds$$
Significance:

• This result is independent of the bootstrap regime and does not require the SG velocity to be Lipschitz.
• It combines the deterministic flow representation for Euler with the superposition principle (Ambrosio-Gigli-Savaré) for SG.
• Corollary: On the bootstrap window, since the velocity difference is $O(\varepsilon)$, the Wasserstein distance between the physical densities is also $O(\varepsilon)$. This provides a direct measure-level stability result.

D. Second-Order Approximation

Result: The paper constructs a first-order corrector $(\rho_1, \phi_1)$ to the Euler solution. By defining a corrected field $\tilde{u}_\varepsilon = \bar{u} + \varepsilon u_1$, the author shows that the residual error between the SG solution and this corrected Euler approximation is $O(\varepsilon^2)$ in velocity.

• This refines the limit, showing that the leading-order error is captured by a linear transport-elliptic system.

4. Significance and Impact

1. Bridging Fluid Dynamics and Optimal Transport: The paper successfully integrates tools from optimal transport (Wasserstein distances, polar factorization) with classical PDE analysis (Calderón-Zygmund, transport equations) to handle the fully nonlinear Monge-Ampère structure of the SG system.
2. Quantitative Rigor: Previous work (e.g., by Loeper) established weak convergence ($\varepsilon^{2/3}$) and local well-posedness. This paper provides strong, quantitative estimates ($O(\varepsilon)$) in strong norms ($L^2$ velocity, $W_2$ density).
3. Extended Validity: The log-log lifespan improvement is a significant theoretical advancement, suggesting that the SG model remains a valid proxy for Euler dynamics for much longer timescales than previously proven, which is crucial for geophysical modeling applications.
4. Robustness: The Wasserstein comparison theorem demonstrates that the stability of the physical densities holds even when the SG velocity field lacks the regularity required for classical Lagrangian flow theory, highlighting the power of the superposition principle in this context.

In summary, Armengou's work provides a comprehensive, quantitative justification for using the Semigeostrophic system as an approximation of 2D Euler in the small-amplitude regime, offering improved time scales and strong stability estimates that were previously out of reach.