Collective dynamics of trail-interacting particles

Paul Pineau, Samuel Bell, Raphaël Voituriez, Ram M. Adar · 2025

Read Collective dynamics of trail-interacting particles by Paul Pineau, Samuel Bell, Raphaël Voituriez, Ram M. Adar (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-2512-03950.txt.

Introduction

Ants leaving scent trails and cells laying chemical tracks share a trick: past paths steer future motion. Paul Pineau, Samuel Bell, Raphaël Voituriez, and Ram M. Adar build a minimal model of many such trail-following particles in a 2025 statistical mechanics preprint. Earlier work treated single particles or averaged theories, leaving collective behaviour largely unexplored.

In their model, particles wander randomly while depositing long-lasting trails that attract or repel those who come later, forming a shared memory field. Using stochastic density functional theory, the authors derive equations for the collective density and study them on paper and in simulation. With repelling trails, densities spread unusually and form temporary clusters. With attracting trails, the system condenses in finite time into frozen, localised clumps. Memory plus noise, they conclude, generates instabilities and self-organisation of its own.

Catalog record

CREATORPaul Pineau, Samuel Bell, Raphaël Voituriez, Ram M. Adar
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-03950

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Collective dynamics of trail-interacting particles Paul Pineau, Samuel Bell, Raphaël Voituriez, Ram M. Adar Abstract Trail interactions occur when past particle trajectories bias future motion, rendering the system out of thermodynamic equilibrium. While such systems are abundant in nature, their understanding is limited to the single-particle level or phenomenological mean-field theories. Here, we introduce a minimal model of many trail-interacting particles that extends this paradigm to the fluctuating collective level. Particles diffuse while depositing long-lasting repelling/attracting trails that act as a shared memory field, coupling their dynamics across time and space. Using stochastic density functional theory, we derive fluctuating hydrodynamic equations and analyze analytically and numerically the resulting behaviors. We show that memory, coupled with fluctuations, fundamentally reshapes collective dynamics; In the repulsive case, the particle density displays superdiffusive spreading characterized by transient clustering and ballistic motion; In the attractive case, the system condensates in finite time into frozen, localized states. Our results establish general principles for trail-interacting systems and reveal how persistent fields generate novel instabilities and self-organization. Many living and synthetic agents modify their environment by depositing persistent trails that bias their subsequent motion. Examples include ants laying pheromone traces Dussutour et al. 2004 ; Jackson et al. 2006 , cells navigating via self-secreted cues Flyvbjerg 2021 ; d’Alessandro et al. 2021 ; Tröger et al. 2024 , and self-propelled droplets guided by trails of micelles Hokmabad et al. 2022 ; Chen et al. 2025 . In all cases, trails introduce a shared memory of past motion that renders these systems out of equilibrium, and drives long-ranged spatial and temporal correlations. Despite the ubiquity of trail-interacting systems, their collective behavior remains poorly understood. Theoretical studies of memory-driven motion have mostly focused on self-interacting random walks, where a single particle’s trajectory is influenced by its own past Toth 1995 ; Pemantle 2007 ; Kranz and Golestanian 2019 ; Brémont et al. 2024 ; Brémont et al. 2025 . Such non-Markovian dynamics result in rich phenomena, including localization, anomalous diffusion, and aging. By contrast, far less is known about the collective behavior of particles coupled through the trails they all generate. As the dynamics depends on the full system’s history, the collective description cannot be reduced to instantaneous pairwise interactions. Existing approaches rely mainly on mean-field or hydrodynamic descriptions Tóth and Werner 2002 ; Adar and Joanny 2024 ; Bell et al. 2025 ; Hartman et al. 2024 ; Maggs 2025 , , as well as numerical simulations Gelimson et al. 2016 ; Meyer and Rieger 2023 ; Meyer et al. 2024 ; Mokhtari et al. 2022 ; Hartman et al. 2024 ; Maggs 2025 . \phi \phi \lambda_{i\to i+1} \lambda_{i\to i-1} x k i Figure 1 : Schematic illustration of the model. Each particle (in red) lives in a potential landscape \phi (in blue). Inset: two possible dynamics of a single particle: hopping to neighboring sites with rates \lambda_{i\pm 1} and trail deposition with rate k . One deposition increase locally \phi by 1 (purple event) . Translucent red dots represents the three possible future state of the particle. In this Letter, we introduce and analyze a minimal model of trail-interacting particles that extends single-particle self-interacting walks to the fluctuating hydrodynamic level. The model couples the motion of diffusing particles to a non-decaying memory field that records their trajectories and biases subsequent motion. Using stochastic density functional theory and large-scale simulations, we show how environment-stored memory reshapes collective dynamics, leading to nontrivial scaling regimes, propagating modes, and condensation phenomena. Our framework establishes general principles for the study of systems with long-lived self-generated fields and opens new directions for exploring memory-driven self-organization in statistical physics. Model. We formulate a lattice model of trail-interacting particles, sketched in Fig. 1 . For simplicity, we focus on the one-dimensional case. We consider N particles distributed along L/a lattice sites, where L is the system size and a the lattice constant. The state of each site i is described by its particle occupancy \rho_{i} and trail content \phi_{i} . The coupled dynamics of both fields, defined for convenience in continuous time t , is driven by two types of independent transition events: (i) hopping of a particle from site i to a neighboring site i\pm 1 , with rate \rho_{i}\lambda_{i\to i\pm 1}=2\rho_{i}k_{tr}/\left(1+\exp\left[H(\phi_{i\pm 1})-H(\phi_{i})\right]\right) and (ii) deposition of a trail unit, with rate \rho_{i}k . Here, k_{tr} is a typical single particle translation rate, related to the diffusion constant D according to D=k_{tr}a^{2}, and H(\phi_{i}) acts as a Hamiltonian that biases the hopping rate according to differences in trail concentration in neighboring sites. Finally, k is a typical trail deposition rate per particle. Note that the trail is not degraded in this model, and most importantly does not diffuse, in contrast to the broad class of models of autochemotactic biological systems Keller and Segel 1970 ; Keller and Segel 1971a ; Keller and Segel 1971b ; Brenner et al. 1998 or autophoretic particles Meyer and Rieger 2023 ; Saha et al. 2014 ; Romano and Gambassi 2025 . It thus acts as a persistent memory field for particle trajectories. The model can be generalized to additional trail dynamics (such as degradation or determinsitic deposition) and higher spatial dimensions in a straightforward manner. The lattice dynamics can be mapped to a stochastic density functional theory (SDFT) in the continuum limit, using exact path-integral methods Andreanov et al. 2006 ; Lefevre and Biroli 2007 ; Saha and Sadhu 2024 ; Mukherjee et al. 2025 . The resulting equations are SM \displaystyle\partial_{t}\rho \displaystyle=D\partial_{xx}\rho+D\partial_{x}\left[\rho\partial_{x}H\left(\phi\right)\right]+\partial_{x}\left(\sqrt{2D\rho}\,\eta\right), \displaystyle\partial_{t}\phi \displaystyle=k\rho+\sqrt{k\rho}\,\xi, (1) and constitute the central result of this paper. The stochastic nature of the dynamics is manifested in noise terms in the particle and trail dynamics, \eta and \xi , respectively. They are both Gaussian white noises with \langle y\rangle=0 and \langle y(t,x)y(t^{\prime},x^{\prime})\rangle=\delta(t-t^{\prime})\delta(x-x^{\prime}) , where y=\eta,\xi , while \langle\xi\eta\rangle=0 . Importantly, the mapping between the lattice dynamics and Eq. ( Collective dynamics of trail-interacting particles ) is exact in the a\to 0 limit, except for the \xi term that approximates the true shot noise of the Poissonian deposition process. This approximation however becomes exact at long times. Despite the apparent complexity and intrinsic non-Markovian nature of trail-mediated interactions, which lie beyond the classical framework of pairwise interactions, Eq. ( Collective dynamics of trail-interacting particles ) shows that the particle density obeys an equation of the Dean-Kawasaki form Dean 1996 ; Kawasaki 1998 with transparent physical interpretation. It expresses the conservation of particles undergoing diffusion and advection according to the effective Hamiltonian H\left(\phi\right) , with multiplicative noise encoding particle number fluctuations. This apparent simplicity – with no explicit interaction term – comes at the cost of a coupling to the additional trail density field \phi , which undergoes a simple irreversible first-order growth kinetics with source term \rho and multiplicative noise originating from the stochastic deposition. We focus below on the example H=h\phi for simplicity. We analyze the linear stability of the system within mean field (MF) and proceed by analyzing correlations beyond MF in both the repulsive and attractive cases ( h>0 and h<0 , respectively). Linear stability analysis and time dependent structure factors. We consider first a simple uniform state with \rho(t=0,x)=\rho_{0} and \phi(t=0,x)=\phi_{0} . Without loss of generality, we set \phi_{0}=0

Read the complete text