[Paper Review] Quantitative Propagation of Chaos for 2D Viscous Vortex Model on the Whole Space

Xuanrui Feng, Zhenfu Wang|arXiv (Cornell University)|Oct 8, 2023

Cosmology and Gravitation TheoriesPhysics and Astronomy3 citations

TL;DR

This paper establishes the first quantitative propagation of chaos estimate for the 2D viscous vortex model on the whole space ℝ², using relative entropy to measure convergence of N-particle systems to their mean-field limit. It derives novel Li-Yau-type and Hamilton-type heat kernel estimates for the 2D Navier-Stokes equation, enabling explicit decay rates in the relative entropy, thus resolving a long-standing open problem in kinetic theory and statistical mechanics of fluid-like particle systems.

ABSTRACT

We derive the quantitative estimates of propagation of chaos for the large interacting particle systems in terms of the relative entropy between the joint law of the particles and the tensorized law of the mean field PDE. We resolve this problem for the first time for the viscous vortex model that approximates 2D Navier-Stokes equation in the vorticity formulation on the whole space. We obtain as key tools the Li-Yau-type estimates and Hamilton-type heat kernel estimates for 2D Navier-Stokes in the whole space.

Motivation & Objective

• To establish a quantitative propagation of chaos result for the 2D viscous vortex model on the whole space ℝ².
• To extend the relative entropy method beyond the torus setting to the non-compact whole space ℝ².
• To derive sharp heat kernel estimates—Li-Yau-type and Hamilton-type—for the 2D Navier-Stokes equation in the whole space.
• To provide explicit, uniform-in-time bounds on the relative entropy between the N-particle joint law and the tensorized mean-field law.
• To resolve the open problem of quantitative chaos propagation for the 2D Navier-Stokes equation in vorticity formulation on ℝ².

Proposed method

• The authors use the relative entropy method to compare the N-particle joint law ρ_N(t) with the tensorized mean-field law ρ̄_t^⊗^N.
• They derive Li-Yau-type gradient estimates and parabolic Harnack inequalities for the 2D Navier-Stokes equation in ℝ² to control the regularity of the mean-field solution.
• Hamilton-type heat kernel estimates are established for the fundamental solution of the 2D Navier-Stokes equation to bound the decay of the density and its derivatives.
• The analysis relies on entropy solutions of the Liouville equation (1.4) and the Fokker-Planck equation (1.3), ensuring well-posedness and integrability.
• Key estimates are derived by differentiating the mean-field equation and applying energy-type inequalities to control terms involving ∇ρ̄, ∇²ρ̄, and ∇³ρ̄.
• The proof proceeds via a hierarchy of differential inequalities for the relative entropy and its higher-order moments, using the structure of the Biot-Savart kernel and the divergence-free property of K.

Experimental results

Research questions

• RQ1 Can a quantitative propagation of chaos estimate be established for the 2D viscous vortex model on the whole space ℝ², where the mean-field limit is governed by the 2D Navier-Stokes equation in vorticity form?
• RQ2 What are the sharp heat kernel estimates—specifically Li-Yau-type and Hamilton-type—for the 2D Navier-Stokes equation on ℝ², and how do they control the regularity of the mean-field solution?
• RQ3 How can the relative entropy between the N-particle joint law and the tensorized mean-field law be controlled uniformly in time and N, especially in the non-compact setting?
• RQ4 What are the necessary and sufficient conditions on the initial data to ensure the relative entropy decays at a rate O(1/N) as N → ∞?
• RQ5 How do the Biot-Savart kernel and the divergence-free structure of the velocity field influence the propagation of chaos in the whole space?

Key findings

• The paper establishes the first quantitative propagation of chaos estimate for the 2D viscous vortex model on ℝ², showing that the scaled relative entropy H_N(ρ_N | ρ̄_N) decays at a rate O(1/N) uniformly in time.
• A novel Li-Yau-type gradient estimate is derived for the 2D Navier-Stokes equation in the whole space, providing control over the logarithmic derivative of the mean-field density.
• Hamilton-type heat kernel estimates are proven for the 2D Navier-Stokes equation on ℝ², enabling precise decay estimates for the fundamental solution and its derivatives.
• The relative entropy satisfies the bound H_N(ρ_N | ρ̄_N)(t) ≤ M e^{Mt} (H_N(ρ_N⁰ | ρ̄_N⁰) + 1/N), where M depends on the initial data, extending previous results from the torus to the whole space.
• The authors prove that the mean-field solution ρ̄_t satisfies higher-order differential inequalities involving |∇ρ̄|²/ρ̄, |∇²ρ̄|²/ρ̄, and (log ρ̄)²ρ̄, which are essential for controlling the entropy dissipation.
• The Biot-Savart kernel K(x) = (1/(2π)) (-x₂, x₁)/|x|² is shown to preserve the structure needed for the entropy method, despite the non-compactness of ℝ².