Lagrangian formulation and Eulerian closure in alignment dynamics

TL;DR

This paper develops a mathematical framework for alignment dynamics, establishing well-posedness, deriving flocking estimates, and analyzing long-term behavior of solutions in both Lagrangian and Eulerian descriptions.

Contribution

It introduces a novel approach to connect Lagrangian and Eulerian formulations, proving asymptotic mono-kinetic closure and stability results for alignment systems.

Findings

Global well-posedness of Lagrangian dynamics

Asymptotic vanishing of defect terms for heavy-tailed interactions

Convergence results towards mono-kinetic Eulerian limits

Abstract

We investigate a continuum Lagrangian $p$-alignment system given by a nonlocal mean-field system of ordinary differential equations for interacting agents with weak initial data. We first establish global well-posedness of the Lagrangian dynamics and derive quantitative flocking estimates. We next construct Eulerian variables from the possibly non-injective Lagrangian flow via pushforward and disintegration, which leads to an Euler--Reynolds--alignment system featuring a nonnegative Reynolds stress and, for $p>2$, a nonlinear defect force induced by microscopic velocity fluctuations. Assuming only heavy-tailed interaction, we then show that these defect terms vanish asymptotically, leading to asymptotic mono-kinetic closure in the long-time limit. In the linear case $p=2$, we further obtain global weak solutions to the Euler--alignment system, including a sharp one-dimensional critical-threshold characterization and a global result in higher dimensions under a large-coupling condition. Finally, we establish a uniform-in-time mean-field stability estimate for the particle Cucker--Smale system in the linear regime and deduce uniform-in-time convergence toward the mono-kinetic Eulerian limit; for general $p\ge2$, we also obtain a finite-time mean-field convergence result toward the associated kinetic/Lagrangian alignment dynamics.