🔬 condensed matter

Disordered Yet Directed: The Emergence of Polar Flocks with Disordered Interactions

This paper demonstrates that in a self-propelled particle model with disordered, spin-glass-like alignment couplings, increasing interaction variance can paradoxically promote global polar order by enabling activity-driven local clustering that mitigates frustration, allowing flocks to emerge even when most interactions are antialigning.

Original authors: Eloise Lardet, Raphaël Voituriez, Silvia Grigolon, Thibault Bertrand

Published 2026-09-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Eloise Lardet, Raphaël Voituriez, Silvia Grigolon, Thibault Bertrand

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the natural world, from the swirling schools of fish to the murmurations of starlings, individual creatures often move as one without a central commander. This phenomenon, known as flocking, has long fascinated scientists who study how simple rules followed by individuals can lead to complex, coordinated group behavior. For decades, researchers have relied on computer models to understand this process, assuming that every member of a group tries to align its direction with its neighbors. These models work well when everyone agrees, but they struggle to explain what happens when the group is full of disagreement or when the rules for getting along are messy and inconsistent. In real life, animals and even microscopic organisms are rarely identical; they vary in their personalities, their senses, and how they react to one another. Understanding how order emerges from such chaos is crucial not just for biology, but for designing swarms of robots or understanding how cells organize themselves in the body.

A team of researchers has now explored this question by building a new kind of computer simulation that deliberately introduces confusion into the mix. Instead of assuming that every particle in their model wants to align perfectly with its neighbors, they assigned each pair of particles a random relationship. Some pairs were programmed to want to face the same way, while others were programmed to want to face opposite directions. In a static system, such as a collection of magnets, this kind of random conflict would typically lead to a frozen, disordered mess where no overall direction emerges. The researchers expected that adding this randomness would destroy the flock. Instead, they discovered a surprising twist: increasing the amount of disagreement actually helped the group organize itself.

The scientists created a virtual world filled with self-propelled particles, tiny entities that move forward on their own, much like bacteria or synthetic robots. They programmed these particles to interact based on a rule where the strength and direction of their influence on each other were drawn from a random distribution. In some cases, the average influence was positive, meaning the group generally wanted to agree. In other cases, the average influence was negative, meaning the group generally wanted to disagree. When the researchers ran their simulations with a low level of randomness, the particles behaved as expected: they either moved together in a unified direction or remained scattered and chaotic. However, as they increased the variance, or the spread, of these random interactions, something unexpected happened. Even when the majority of the interactions were set to push the particles apart, the group began to form a cohesive flock.

This counterintuitive result relies on the fact that the particles are active; they are constantly moving and exploring their environment. In a static system, particles are stuck with their neighbors and cannot escape the frustration of conflicting instructions. But because these particles are moving, they can physically rearrange themselves. The simulation showed that the particles naturally drifted toward neighbors with whom they shared a strong, positive connection. They effectively sorted themselves out, clustering together with the few partners that helped them align, while drifting away from those that caused conflict. This self-organization allowed them to build local pockets of agreement. Once these small, aligned clusters formed, the activity of the particles helped these clusters grow and connect, eventually leading to a state where the entire group moved in the same direction, despite the fact that most of the underlying rules were actually trying to prevent it.

The researchers tested this idea rigorously, running thousands of simulations with different numbers of particles, different densities, and different levels of noise. They found that this effect was robust. It occurred even when the average interaction was strongly negative, meaning the system was designed to fail. They also checked what would happen if the random rules were not fixed but could change over time, a scenario known as annealed disorder. In that case, the flocking did not happen. This confirmed that the key was the combination of fixed, random rules and the ability of the particles to move. The particles needed a stable set of relationships to learn which neighbors were helpful and which were not, and then use their movement to seek out the helpful ones.

The study also revealed that this process creates a specific kind of structure within the group. The particles did not just form a uniform cloud; they formed dense bands and clusters where the local alignment was very strong. These clusters were not perfect, and a significant portion of the interactions remained frustrating, but the system managed to find a balance where the overall movement was directed. The researchers observed that as the randomness increased, the particles spent more time in contact with their aligned neighbors, effectively filtering out the noise. This suggests that in real-world systems, where variability is the norm rather than the exception, disorder might not be an obstacle to coordination but a mechanism that drives it.

The findings challenge the long-held assumption that uniformity is necessary for collective motion. In many previous models, introducing heterogeneity or disagreement was thought to weaken or destroy the flock. This new work suggests that in active systems, where individuals are constantly moving and interacting, a certain amount of disorder can actually be beneficial. It allows the system to self-sort, finding the most cooperative pathways through the chaos. The researchers noted that while their results were derived from computer simulations, they offer a new perspective on how natural swarms, from bacterial colonies to bird flocks, might maintain order despite the inherent differences between individuals. The ability to turn a sea of conflicting signals into a unified direction appears to be a fundamental property of active matter, driven by the simple, physical act of moving toward what works and away from what does not.

Technical Summary: Disordered Yet Directed: The Emergence of Polar Flocks with Disordered Interactions

Problem Statement
Collective flocking behavior, typically modeled by the Vicsek model, arises from simple local alignment rules where self-propelled particles (SPPs) align their velocities with neighbors. While the robustness of flocking to various perturbations (e.g., noise, obstacles) has been extensively studied, the impact of quenched disorder in the interaction couplings themselves remains underexplored. In equilibrium spin systems, such disorder (as seen in spin-glass models) typically induces frustration, preventing long-range order and leading to a glassy phase with only local correlations. The central question addressed is whether introducing random, quenched disorder into the pairwise alignment couplings of an active system destroys flocking or, paradoxically, facilitates it.

Methodology
The authors introduce the Active Disordered XY (ADXY) model, an extension of the Vicsek model where the coupling strength $K_{ij}$ between particle pairs is a quenched random variable drawn from a Gaussian distribution $\mathcal{N}(K, \sigma_K^2)$ .

Dynamics: The system consists of $N$ pointlike particles moving at constant speed $v_0$ . Their orientation $\theta_i$ evolves according to an overdamped Langevin equation:
$\dot{\theta}_i = \frac{1}{n_i} \sum_{j \in N_i} K_{ij} \sin(\theta_j - \theta_i) + \eta \xi_i$
where $n_i$ is the number of neighbors within an interaction radius $\sigma_I$ , $\eta$ is the noise strength, and $\xi_i$ is Gaussian white noise.
Couplings: $K_{ij} > 0$ represents ferromagnetic (aligning) interactions, while $K_{ij} < 0$ represents antiferromagnetic (antialigning) interactions. The couplings are fixed in time (quenched) but spatially random.
Simulation: The authors perform large-scale numerical simulations (up to $N \approx 10^5$ ) using the Euler-Maruyama method. They analyze the global polar order parameter $\Psi$ , demixing order parameter $d$ , and cluster size $r_c$ across varying mean coupling $K$ , coupling variance $\sigma_K$ , and density $\rho$ .
Comparisons: The study contrasts the quenched disorder model with an "annealed" variant (where couplings update dynamically when neighbors change) and validates results using bimodal coupling distributions and topological (k-nearest neighbor) interactions.

Key Contributions and Results

Paradoxical Promotion of Order by Disorder:
Contrary to equilibrium expectations where disorder destroys order, increasing the variance $\sigma_K$ of the coupling distribution promotes global polar order. Remarkably, the system exhibits a flocking phase even when the mean coupling $K$ is negative (i.e., when the majority of interactions are antialigning). As $\sigma_K$ increases, the critical value $K_c$ for the onset of flocking shifts significantly into the negative regime.
Mechanism of Self-Organization:
The emergence of order in the presence of negative mean coupling is driven by a self-sorting mechanism. Activity allows particles to dynamically rearrange into local clusters where they happen to be surrounded by neighbors with strong positive (ferromagnetic) couplings.
- Clustering: As $\sigma_K$ increases, the distribution of neighbors becomes skewed, and particles form cohesive clusters of highly aligning individuals.
- Frustration Reduction: While the system retains inherent frustration (unfavorable interactions), activity reduces the global frustration by allowing particles to escape unfavorable neighborhoods and settle into favorable ones.
- Role of Quenched Disorder: This mechanism relies specifically on quenched disorder. In the annealed variant, where couplings update dynamically, the system fails to achieve global order despite forming local clusters, indicating that the fixed nature of the disorder is essential for sustaining the effective interaction network required for long-range order.
Phase Diagram and Transitions:
- The authors map a phase diagram in the $(K, \rho)$ plane. For large $\sigma_K$ , a region of global polar order exists where $K \leq 0$ .
- The transition to order is accompanied by a sharp increase in particle persistence and pairwise contact times.
- A "banding" phase (ordered traveling bands) is observed between the disordered and homogeneous polar liquid phases, similar to the original Vicsek model but subject to stronger finite-size effects.
- The transition appears to be first-order (supported by Binder cumulant analysis for small $\sigma_K$ ), and the ordered phase retains true long-range order with algebraic decay consistent with Toner-Tu theory ( $\Psi - \Psi_\infty \sim L^{-2/3}$ ).
Robustness and Fragility:
- The results are robust across different interaction distributions (bimodal) and interaction types (topological vs. metric).
- However, flocks formed under quenched disorder are found to be more fragile than those in the standard Vicsek model ( $\sigma_K=0$ ), exhibiting larger angular diffusion coefficients and requiring larger system sizes to stabilize banding.

Significance and Claims
The paper claims to provide a general mechanism for how global order can emerge in active systems with heterogeneous, randomly distributed interactions—a scenario common in natural and synthetic swarms (e.g., cellular tissues, bacterial suspensions). The primary finding is that activity acts as a key ingredient to reduce frustration in disordered systems. By enabling spatial self-organization, activity allows particles to "select" neighborhoods with favorable interactions, thereby overcoming the frustration that would otherwise prevent flocking in a static or equilibrium system.

The authors emphasize that this counterintuitive result—disorder promoting order—relies on the specific interplay of reciprocal quenched couplings and active dynamics. They note that while local clustering is necessary, it is insufficient for global order without the specific constraints imposed by quenched disorder. The work suggests that variability and heterogeneity in alignment interactions should be viewed not merely as noise to be overcome, but potentially as a feature that, when coupled with activity, can facilitate collective motion even in regimes where mean-field interactions would predict disorder. Future work is suggested to explore the analytical challenges of combining active matter theory with spin-glass theory to fully characterize these frustrated active phases.

More like this