System size and boundaries determine the patterning dynamics of attracting active particles

Jan Rombouts, Michael L Zhao, Alexander Aulehla, Anna Erzberger · 2025

Read System size and boundaries determine the patterning dynamics of attracting active particles by Jan Rombouts, Michael L Zhao, Alexander Aulehla, Anna Erzberger (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-2509-08533.txt.

Introduction

Jan Rombouts and 3 co-authors find that patterns in self-attracting particles slow down sharply as their box grows in a 2025 paper. Pattern growth inside walls is common, from cells to robot swarms. Yet how walls steer the timing was not clear. The team works with equations that mix drift and spread, with a pull that acts at range. Such equations catch the shared motion of active particles that attract each other.

As box size crosses set marks, slow modes appear in steps. Pattern times then grow at an exponential rate. Tests on living cell groups match the forecast. Organs have fixed size as they grow their patterns. Size alone may set the clock of development more than genes do, a thought this work makes testable.

Catalog record

CREATORJan Rombouts, Michael L Zhao, Alexander Aulehla, Anna Erzberger
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-08533

Browse the full library · About the Tower of Babel

Full text

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

System size and boundaries determine the patterning dynamics of attracting active particles Jan Rombouts, Michael L Zhao, Alexander Aulehla, Anna Erzberger Abstract Pattern formation often occurs in confined systems, yet how boundaries shape patterning dynamics is unclear. We develop techniques to analyze confinement effects in nonlocal advection-diffusion equations, which generically capture the collective dynamics of active self-attracting particles. We identify a sequence of size-controlled transitions that generate characteristic slow modes, leading to exponential increase of patterning timescales. Experimental measurements of multicellular dynamics confirm our predictions. Introduction. Attractive interactions between the constituents of active systems lead to rich collective phenomena, ranging from assemblies of active colloids Theurkauff et al. 2012 ; Palacci et al. 2013 to the formation of animal groups Vicsek and Zafeiris 2012 ; Ioannou and Laskowski 2023 . In multicellular systems, pattern formation through cell aggregation appears in many different situations, from the formation of slugs in Dictyostelium Weijer 2004 to the recruitment of immune cells to a site of infection Strickland et al. 2024 . Self-organized aggregation is central to development Ros-Rocher et al. 2023 ; Abitua et al. 2024 , synthetic patterning McNamara et al. 2024 ; Fiuza et al. 2024 ; Toda et al. 2018 , and possibly the evolution of multicellularity ( Brunet and King 2017 ) . At larger spatial scales, interactions between individual organisms lead to the collective structures such as fish schools, bird flocks or human crowds. Although these processes often take place in confined environments, how boundary interactions and geometrical constraints affect aggregation-induced pattern formation is not well understood. Nonlocal advection-diffusion models offer a powerful framework for describing the dynamics of self-attracting constituents without requiring detailed knowledge of interaction mechanisms. By coarse-graining molecular, chemical, or mechanical interactions into an effective kernel, they can capture a broad class of physical and biological systems Painter et al. 2024 . While such models have been applied from subcellular to ecological scales, most studies focus on infinite or periodic domains (but see Refs. Hillen and Buttenschön 2020 ; Yu et al. 2025 ), and direct comparisons with experimental data remain rare. In this Letter, we investigate how confinement alters pattern formation in a minimal nonlocal model of attractive active particles. We show that increasing system size triggers a sequence of bifurcations that generate slow dynamical modes, leading to an exponential increase of the patterning timescale. These theoretical predictions are confirmed by experimental data of primary reaggregated murine cells which form patterns on size-controlled substrates. Our results reveal how finite system size and boundary interactions control the timescale of pattern formation driven by long-range attraction. Non-local advection-diffusion dynamics. We investigate the influence of finite system size and boundaries on the collective dynamics of self-attracting particles whose density field \rho(\mathbf{r},t) evolves according to the mass-conserving advection-diffusion equation \partial_{t}\rho=\nabla\cdot\left(D\nabla\rho-\mathbf{v}(\mathbf{r})\rho\right). (1) The particles undergo random motion set by the diffusion coefficient D , and advective motion with velocity \mathbf{v}(\mathbf{r}) (Fig. 1 a). We focus on flows driven by forces arising due to interactions between particles and with the boundaries, such that in the overdamped limit, the velocity field satisfies \gamma(\mathbf{r})\mathbf{v}(\mathbf{r})=\mathbf{f}_{\rm{p}}(\mathbf{r})+\mathbf{f}_{\rm{b}}(\mathbf{r}), (2) in which we introduce \mathbf{f}_{\rm{p}}(\mathbf{r}) and \mathbf{f}_{\rm{b}}(\mathbf{r}) as the forces acting at \mathbf{r} due to particle-particle and particle-boundary interactions respectively. The dissipative coefficient \gamma(\mathbf{r}) is taken to depend on the local density, \gamma(\mathbf{r})=\gamma(\rho) , such that higher densities impede particle movement. Here, we use \gamma(\rho)=\gamma_{0}/(1-\rho/\rho_{\rm{max}}) , an expression derived from microscopic random walks with occupation-dependent jumping probabilities ( Painter and Hillen 2002 ) . We represent generic interaction forces using non-local kernels of the form \mathbf{f}_{\rm{p}}(\mathbf{r})=\int_{\Omega}\mathbf{G}_{\textup{p}}(\mathbf{r^{\prime}}-\mathbf{r})\rho(\mathbf{r}^{\prime})\mathrm{d}\mathbf{r^{\prime}},\quad\mathbf{f}_{\rm{b}}(\mathbf{r})=\int_{\Omega^{c}}\mathbf{G}_{\textup{b}}(\mathbf{r}^{\prime}-\mathbf{r})\mathrm{d}\mathbf{r^{\prime}}, (3) in which \Omega is the domain containing the particles and \Omega^{\rm{c}} is the space outside (Fig. 1 a,b). More generally, a boundary interaction can be expressed as \int\mathbf{G}_{\textup{b}}(\mathbf{r}^{\prime}-\mathbf{r})\phi(\mathbf{r}^{\prime})\mathrm{d}\mathbf{r^{\prime}} , where \phi denotes a phase field defining the external material. Eq. ( 3 ) is obtained for \phi=1 outside the domain and \phi=0 inside. The interaction kernels are parameterized as \mathbf{G}_{i}(\mathbf{r}^{\prime}-\mathbf{r})=\frac{\chi_{i}}{\sigma_{i}}g_{i}\left(\frac{|\mathbf{r}^{\prime}-\mathbf{r}|}{\sigma_{i}}\right)\frac{\mathbf{r}^{\prime}-\mathbf{r}}{|\mathbf{r}^{\prime}-\mathbf{r}|} (4) such that \sigma_{i} and \chi_{i} determine the length scale and amplitude of the interaction, and the normalized function g_{i} describes how the magnitude of the force depends on distance, with i\in\{\rm{p};\rm{b}\} (Fig. 1 b). Different microscopic mechanisms can be shown to map to such interaction kernels. For example, Keller-Segel-type chemotactic attraction between motile cells ( Keller and Segel 1970 ; Painter 2019 ) leads to an exponentially decaying interaction kernel with length scale \sigma_{\textup{p}}=\sqrt{D_{c}/b} , determined by the diffusion constant D_{c} and the degradation rate b of the chemoattractant ( Lee et al. 2001 ) . Similarly, for contractile fluids ( Bois et al. 2011 ; Palmquist et al. 2022 ) , the equivalent nonlocal description contains an exponential kernel with decay length corresponding to the hydrodynamic lengthscale \sqrt{\eta/\gamma} , with \eta the viscosity and \gamma a friction coefficient sup . Figure 1: Pattern formation of self-attracting particles in finite-sized systems. In a one-dimensional system of length \ell , active particles (orange) undergo random and directed motion due to interparticle coupling \alpha and particle-boundary coupling \beta . b) We describe interparticle and particle-boundary interactions with nonlocal exponential kernels with characteristic decay length \sigma . c) Simulations of Eq. ( 5 ) illustrate the patterning dynamics for two different domain lengths. ( \alpha=18,\beta=7.2,\rho_{0}=0.4 ) Considering interaction kernels with equal shape and range for the inter-particle and boundary-particle interactions in a one-dimensional system, we obtain the equation \displaystyle\partial_{t}\rho=\partial_{x}\biggl(\partial_{x}\rho-\rho(1-\rho)\biggl[ \displaystyle\alpha\int_{0}^{\ell}\tilde{g}(x^{\prime}-x)\rho(x^{\prime})\mathrm{d}x^{\prime} (5) \displaystyle+ \displaystyle\beta\int_{\mathbb{R}\backslash[0,\ell]}\tilde{g}(x^{\prime}-x)\mathrm{d}x^{\prime}\biggl]\biggr), in which space, time, and the density have been rescaled by \sigma,\tau=\sigma^{2}/D and \rho_{\text{max}} respectively sup , the system size is given relative to the interaction range \ell=L/\sigma , and \tilde{g}(x^{\prime}-x)=\sign(x^{\prime}-x)g(|x^{\prime}-x|) , and the external material is assumed constant in time. The dynamics are controlled by the dimensionless particle-particle interaction parameter \alpha=\chi_{\textup{p}}\rho_{\text{max}}\sigma/(\gamma_{0}D) and the particle-boundary interaction parameter \beta=\chi_{b}\sigma/(\gamma_{0}D) , as well as the total number of particles, which determines \rho_{0}=\big(\int_{0}^{\ell}\rho(x)\mathrm{d}x\big)/\ell . Fig. 1 c shows simulations of Eq. ( 5 ) with no-flux boundary conditions, illustrating how spontaneous patterns form in this system. Conditions for patterning On periodic or infinite domains, patterns form spontaneously from a homogeneous particle distribution if sup ; Jewell et al. 2023 \alpha\rho_{0}(1-\rho_{0})>\frac{1}{2}. (6) Thus, since \alpha represents the relative strength of attraction compared to diffusion, patterning is possible if attraction overcomes random motion and if the average density is neither too low nor too high. While all uniform profiles \rho(x)=\rho_{0} are steady states on infinite domains, this is not the case on bounded domains, for which instability thresholds are analytically accessible only in specific limits. In particular, considering a neutral boundary interaction with \beta=\alpha\rho_{0} allows linearizing Eq. ( 5 ) around a uniform steady state \rho(x)=\rho_{0} , leading to the eigenvalue problem sup (\partial_{xx}-\bar{\alpha}\partial_{x}\mathcal{A})u=\omega u,\qquad(\partial_{x}u-\bar{\alpha}\mathcal{A}u)|_{x\in\{0;\ell\}}=0 (7) in which \bar{\alpha}=\alpha\rho_{0}(1-\rho_{0})\quad\text{and}\quad(\mathcal{A}u)(x)=\int_{0}^{\ell}\tilde{g}(x^{\prime}-x)u(x^{\prime})\mathrm{d}x^{\prime}. (8) The solutions of this equation yield the eigenfunctions u and associated growth rates \omega (Fig. 2 ). In contrast to infinite or periodic domains, these spatial modes are in general not sinusoidal: their shape depends on the domain length and the interaction parameters (Fig. 2 b). Figure 2: Linear stability analysis of the nonlocal equation in finite-sized systems. a) Approximations (dotted lines, Eq. 9 ) and exact values (solid lines, numerical solutions of Eq. 6 in sup ) of the eigenvalues \omega_{n} show that eigenvalue curves (‘loci’) undergo turning accompanied by the cross-over of

Read the complete text