The Mean-Field Ott-Antonsen Manifold is an Unstable Manifold in the Continuum Limit
Christian Kuehn, Giacomo Landi · 2025
Read The Mean-Field Ott-Antonsen Manifold is an Unstable Manifold in the Continuum Limit by Christian Kuehn, Giacomo Landi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-03833.txt.
Introduction
Christian Kuehn and Giacomo Landi compare two continuum accounts of Kuramoto-type interacting particle systems. The continuum limit tracks particles pointwise, while the mean-field limit follows a typical particle, and the two equations differ in structure despite both assuming infinitely many particles.
Solutions of one equation can be generated from solutions of the other, the authors first show. Their main result then proves that the unstable manifold of the homogeneous steady state in the continuum limit is the direct analogue of the famous Ott-Antonsen manifold in the mean-field limit. That direct link clarifies a long-standing puzzle about how the two reduced accounts relate in time.
Catalog record
| CREATOR | Christian Kuehn, Giacomo Landi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-03833
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The Mean-Field Ott-Antonsen Manifold is an Unstable Manifold in the Continuum Limit Christian Kuehn, Giacomo Landi Abstract We study interacting particle systems of Kuramoto-type. Our focus is on the dynamical relation between the partial differential equation (PDE) arising in the continuum limit (CL) and the one obtained in the mean-field limit (MFL). Both equations arise when we are considering the limit of infinitely many interacting particles but the classes of PDEs are structurally different. The CL tracks particles effectively pointwise, while the MFL is an evolution for a typical particle. First, we briefly discuss the relation between solutions of the CL and the MFL showing how to generate solutions of the CL starting from solutions of the MFL. Our main result concerns a dynamical relation between important invariant manifolds of the CFL and the MFL. In particular, we give an explicit proof that the unstable manifold of the homogeneous steady state of the CL is the direct dynamical analogue of the famous Ott-Antonsen manifold for the MFL. 1 Introduction Interacting particle systems form a broad class of coupled differential equations used to describe populations of interacting agents (“particles”) that can display collective behavior without any centralized control. Such systems are employed across a wide range of scientific disciplines to model diverse phenomena, including opinion dynamics [ 19 ] , collective organization in animal groups [ 3 ] , [ 5 ] , [ 32 ] , synchronization of coupled oscillators [ 26 ] , and biological tissue formation [ 6 ] . These models have also attracted considerable attention in the mathematical community, especially in the context of Kuramoto-type systems [ 2 , 18 ] , with particular focus on their behavior in the large-population (infinite-particle) limit, which is the main subject of this work. The relevance of studying large-population limits becomes clear when considering the size of certain systems of interest. For example, starling flocks can contain several thousand individuals [ 5 ] , while mammalian neural networks may involve up to 10^{11} neurons [ 21 ] . Simulating such massive populations by directly integrating a system of N coupled equations, with potentially N^{2} pairwise interactions, quickly becomes computationally prohibitive as N grows. To overcome this challenge, a key idea, introduced in the seminal works [ 9 ] , [ 13 ] , [ 17 ] , [ 22 ] , [ 39 ] , [ 48 ] , [ 50 ] , is to shift the focus from the individual particle trajectories to the evolution of the population’s density, leading to a mean-field description of the system. Mean-field limit of exchangeable particle systems. For each particle, we fix a phase space \mathcal{X}\subseteq\mathbb{R}^{d} with d\in\mathbb{N} . The configuration of N interacting particles is denoted by (x_{i}^{N})_{i\in\{1,\cdots,N\}} and evolves according to this system of coupled ordinary differential equations: \begin{cases}\dot{x}^{N}_{i}(t)=\frac{1}{N}\sum_{j=1}^{N}\phi(x_{i}^{N}(t),x_{j}^{N}(t))\qquad\forall i=1,\ldots,N,\quad t\in[0,T]\\ x^{N}_{i}(0)=x^{N,0}_{i}\qquad\forall i=1,\ldots,N,\end{cases} (DS) where • the index set \{1,\ldots,N\} labels the individual particles, characterizing their identities; • (x_{i}^{N})_{i\in\{1,\cdots,N\}} represents the state of the interacting particles, and may represent positions [ 15 ] , opinions [ 19 ] , velocities [ 12 ] , phases [ 26 ] , or angles [ 53 ] . The model then determines the dimension d of the state space; • \phi:\mathbb{R}^{d}\times\mathbb{R}^{d}\rightarrow\mathbb{R}^{d} encodes the pairwise interaction between particles. We will refer to it as the interaction function of the system. Under standard assumptions on \phi , the solution (x_{i}^{N})_{i\in\{1,\cdots,N\}} of the system exists in C([0,T];(\mathbb{R}^{d})^{N}) . As N\rightarrow+\infty one can show that the system ( DS ) is well approximated by its so-called mean-field limit \mu\in C([0,T];\mathcal{P}(\mathbb{R}^{d})) , which satisfies the following equation: \begin{cases}\partial_{t}\mu_{t}(x)+\nabla_{x}\cdot\left(\mu_{t}(x)\int_{\mathbb{R}^{d}}\phi(x,y)\text{d}\mu_{t}(y)\right)=0,\\ \mu_{t=0}=\mu_{0}.\end{cases} (MFL) commonly referred to as the Vlasov equation in the context of interacting particle systems (IPS). When the interaction function \phi is a gradient, the equation is known in the literature as the aggregation equation. The interpretation of \mu is usually the following one: let \Delta x\subset\mathbb{R}^{d} , then \mu_{t}(\Delta x) represents the probability of finding an agent with position in \Delta x at time t . The relationship between the particle system ( DS ) and its mean-field limit equation ( MFL ) is elucidated by introducing the empirical measure \mu_{t}^{N}\coloneqq\frac{1}{N}\sum_{i=1}^{N}\delta_{x_{i}^{N}(t)} which can be shown to satisfy the mean-field equation ( MFL ), see [ 17 , Th. 1.3.1] . The convergence of the microscopic dynamics (x_{i}^{N})_{i=1}^{N} governed by ( DS ) towards the mean-field solution \mu of ( MFL ) can be derived using a stability argument for the mean-field equation; see [ 17 , 13 ] . This argument yields an estimate of the form: W_{1}(\mu_{t},\mu_{t}^{N})\leq C(T)W_{1}(\mu_{0},\mu_{0}^{N}), where W_{1} is the 1-Wasserstein distance [ 13 ] . See also [ 31 ] where Wasserstein distances with different exponents different from 1 are used in the same context. For a detailed introduction on Wasserstein distances, we refer the reader to [ 54 ] . The continuum limit. The previously discussed mean-field limit establishes the weak convergence of the microscopic dynamics ( DS ) towards the solution of the transport equation ( MFL ). An alternative approach, known as the continuum limit, instead yields pointwise convergence of the solutions of ( DS ) to those of an integro-differential equation: \begin{cases}\partial_{t}x(t,\xi)=\int_{0}^{1}\phi\big(x(t,\xi),x(t,z)\big)\textnormal{d}z,\\ x(0,\xi)=x_{0}(\xi)\end{cases} (CL) where, in this case, the solution x(t,\xi) represents the position in \mathbb{R}^{d} at time of the particle labeled with \xi , where \xi\in[0,1] is a continuous representation of particle’s label. This limit is as important as the previous one, especially in the context of network dynamics and graph limits, two research fields that have seen significant development in recent years, see for example [ 37 , 38 , 23 , 16 ] . Remark 1.1 . The general case of non-exchangeable particles, i.e., when the interaction function is of the form \phi=\phi(\xi,\zeta,x,y) is also of big interest, see for example [ 24 ] , [ 37 ] and [ 4 ] . One important and remarkable example is the Kuramoto model \dot{\theta}_{i}=\omega_{i}+\frac{K}{N}\sum_{j=1}^{N}\sin(\theta_{j}-\theta_{i}),\qquad\forall i=1,\ldots,N where the \omega_{i} represent the natural frequencies of the oscillators, K the coupling strength between the oscillators and \theta_{i}\in\mathbb{S}^{1} is the phase of oscillator i , see [ 2 ] , [ 45 ] and [ 18 ] for a broader and complete perspective about this model. Here, the non-exchangeability is encoded in the different natural frequencies \omega_{i} . Understanding the relations and the properties of the ( CL ) and the ( MFL ) is as important and challenging as there are no previous results available that directly link important dynamically invariant sets in both PDE limits of IPS. The goal of this paper is to deepen the connection between the MFL and the CL PDEs. In particular, we will show how to obtain solutions of CL starting from solutions of MFL . The second and main result of the paper is the explicit form of a family of unstable manifolds for the CL obtained applying the previous transformation to the Ott-Antonsen (OA) manifold of the MFL . The OA manifold has been discovered in the seminal paper [ 41 ] and since then has been widely studied and discussed, see for example [ 52 ] , [ 42 ] , [ 14 ] and [ 33 ] . It provides a