Spatially Structured Cohesion from Extremal Alignment in Topological Active Matter

1. Problem Statement

Active matter systems, such as bird flocks or bacterial swarms, rely on alignment interactions to achieve collective motion. Standard theoretical models (e.g., the Vicsek model) implement alignment as a relaxational process where agents continuously adjust their velocity toward the local average orientation of their neighbors.

• The Limitation: In unbounded (open) systems, this relaxational alignment is effectively decoupled from local density. Without explicit attractive forces (e.g., attraction to a center of mass) or periodic boundary conditions, alignment alone cannot sustain self-confined collective motion; the group tends to disperse.
• The Gap: While natural agents often make discrete decisions rather than continuous relaxations, previous attempts to model "decision-based" active matter (specifically "predictive alignment") resulted in unrealistic, ultra-dense states where the system collapses into a mean-field regime, suppressing genuine spatial structure.
• The Question: Can self-confinement and spatially structured collective motion emerge directly from the alignment rule itself, without explicit attractive forces, by incorporating biologically motivated constraints?

2. Methodology

The authors propose a scanning alignment model based on extremal (arg max) decision rules combined with topological interactions.

• Agent Dynamics:

  • $N$ self-propelled particles move in 2D open space with constant speed $v_0$.
  • At each time step, an agent selects a discrete reorientation angle $\alpha$ from a set $\{-\alpha_d, 0, \alpha_d\}$.
  • The selection maximizes a decision utility function $U_i(\alpha)$.

• The Utility Function:
  The utility is defined as the product of the number of recruited neighbors and the average alignment score:
  $$U_i(\alpha) = N_i(\alpha) \cdot S_i(\alpha)$$
  Where:

  • $N_i(\alpha)$ is the number of neighbors within the agent's vision cone if it were to adopt the candidate orientation $\alpha$.
  • $S_i(\alpha)$ is the average alignment score (dot product of candidate velocity and neighbor velocities) with those neighbors.

• Key Distinctions (Rigid vs. Scanning):

  • Rigid Agents: The interaction neighborhood is fixed based on the current orientation. The utility maximization reduces to standard topological alignment, which fails to produce confinement in open space.
  • Scanning Agents (The Core Model): The interaction neighborhood is re-evaluated for every candidate orientation. The agent "scans" potential future headings to see which direction captures the most neighbors while maintaining alignment. This couples orientational decisions to local density.

• Constraints:

  • Topological Interactions: Agents interact with the $k$ nearest neighbors within their vision cone (rather than all neighbors within a metric radius).
  • Finite Vision Cone: Agents have a limited field of view (opening angle $\beta$).

• Simulation Parameters:

  • System size $N=200$ (verified up to 600).
  • Control parameters: Noise intensity ($\eta$), reorientation step ($\alpha_d$), vision cone angle ($\beta$), and topological neighbor number ($k$).

3. Key Contributions

1. Mechanism for Self-Confinement: The paper identifies candidate-dependent extremal alignment as a minimal mechanism for generating cohesive, spatially structured active matter without explicit attractive forces. The multiplicative structure of the utility function ($N_i \times S_i$) creates an effective cohesive bias: agents prefer directions that align them with more neighbors.
2. Role of Topology: The authors demonstrate that topological interactions (fixing the number of neighbors $k$) are crucial. In metric models, the same extremal rule causes unphysical collapse (infinite density). Topological constraints saturate the cohesive bias, stabilizing finite-sized flocks.
3. Rich Phase Diagram: The model generates a diverse dynamical phase diagram including polarized flocks, cohesive swarms, and persistent swirling states, depending on noise and turning rates.

4. Key Results

A. Parameter Optimization

• Neighbor Number ($k$): The system size decreases as $k$ increases but saturates around $k \approx 7$. This value aligns with empirical findings in starling flocks (Ballerini et al., 2008).
• Vision Cone ($\beta$):
  • Narrow cones ($\beta = 5\pi/12$): Support "hunter-type" dynamics, leading to V-shaped flocks and strong rotational states (swirls).
  • Broad cones ($\beta = 7\pi/12$): Support "prey-type" dynamics, leading to circular flocks and suppressing rotation.

B. Stability and Confinement

• Scanning vs. Rigid: Only scanning agents exhibit genuine stationary confinement ($\gamma \approx 0$ in the scaling of system size $\delta_{c.m.} \sim t^\gamma$) in open space. Rigid agents always disperse ($\gamma > 0$).
• Stability Boundaries: Confinement persists for noise intensities below a threshold that increases with the reorientation angle $\alpha_d$.

C. Dynamical Phases

The system exhibits three main phases within the stable regime:

1. Polarized Flocks: Low noise, high alignment. Agents move coherently in a specific direction.
   • Narrow cones: Can form V-shapes due to anisotropic sampling.
   • Broad cones: Form circular flocks.
2. Swirls (Rotational States): Intermediate noise and large reorientation angles (specifically for narrow cones). The group undergoes persistent collective rotation around the center of mass.
   • Characterized by non-zero angular velocity $|\omega| > 0$.
   • Exhibits chirality switching: The group spontaneously switches between clockwise and counter-clockwise rotation.
3. Cohesive Swarms: High noise, loss of global orientational order ($\phi \approx 0$), but the group remains spatially confined ($\gamma \approx 0$).

D. Correlation and Switching

• Velocity Correlations: Swirling states show damped oscillations in velocity autocorrelation, reflecting rotational memory.
• Chirality Switching: In the swirling phase, the system behaves like a particle in a double-well potential (metastable chiral states). As noise increases, the barrier between clockwise and counter-clockwise states lowers, increasing the switching rate until the rotational state collapses.

5. Significance

• Theoretical Advancement: This work challenges the standard paradigm that cohesion in active matter requires explicit attractive forces. It demonstrates that decision-making logic (specifically, evaluating neighbors based on candidate actions) is sufficient to generate effective attraction.
• Biological Relevance: The model bridges the gap between simple alignment rules and complex biological behaviors. The emergence of $k \approx 7$ and the specific types of collective motion (swirls, V-shapes) mirror observations in real animal groups (e.g., starlings).
• Conceptual Shift: It reframes active matter from a physics of "forces" to a physics of "decisions." The mechanism is interpreted as a "copy-the-group" or majority-following strategy, linking active matter to social dynamics and conformity.
• Topological Regularization: It resolves the issue of density collapse found in previous metric-based extremal models, showing that topological constraints are essential for realistic, spatially structured collective motion.

In conclusion, the paper establishes that candidate-dependent extremal alignment, regularized by topological interactions, is a robust, minimal mechanism for generating self-confined, spatially structured active matter in open environments.