Criticality and Phase Transitions

Swarm behavior transitions sharply across control-parameter values — density, noise, coupling strength — and those transitions have the same statistical signatures as phase transitions in equilibrium physics. This page is the minimum shared vocabulary our phase-transition paper expects readers to carry in.

The Vicsek model

The Vicsek model (Vicsek, Czirók, Ben-Jacob, Cohen, Shochet, 1995) is the minimal continuous-time model of flocking. Each agent updates its heading to the average heading of neighbors within a radius, plus Gaussian noise. The model exhibits a noise-driven phase transition between two regimes:

Low noise — collective motion; agents flow coherently.
High noise — disordered; headings are effectively random.

The transition looks first-order in some variants and second-order in others; the debate drove much of the 2000s active-matter literature. The Vicsek universality class is the benchmark to measure new flocking variants against.

Order parameter

An observable whose mean value distinguishes the two phases. For flocking the canonical choice is alignment:

\psi = \frac{1}{N} \left\lVert \sum_k \frac{\vec v_k}{\lVert \vec v_k \rVert} \right\rVert

which is 0 in the fully disordered phase and 1 when every agent moves in the same direction. Implemented as gossamer.utils.metrics.alignment.

Susceptibility

Variance of the order parameter, scaled by system size:

\chi = N \cdot \mathrm{Var}(\psi).

Peaks sharply at the critical point. One of the strongest empirical signatures of a phase transition — if you can’t produce a susceptibility peak, you probably don’t have a real transition. Implemented as gossamer.metrics.criticality.susceptibility.

Binder cumulant

U = 1 - \frac{\langle \psi^4 \rangle}{3 \langle \psi^2 \rangle^2}.

U depends only weakly on system size at the critical point, but strongly away from it. Plotting U against the control parameter for several system sizes produces a curve crossing at T_c — the cleanest finite-size scaling estimator available. Implemented as gossamer.metrics.criticality.binder_cumulant.

Correlation length

Extracted from the velocity correlation function

C(r) = \left\langle \delta \vec v_i \cdot \delta \vec v_j \right\rangle_{|r_i - r_j| \sim r}

(fluctuations around the mean velocity). The first zero-crossing of C(r) is the conventional proxy for the correlation length ξ. ξ diverges at the critical point; its scaling with system size gives you one of the critical exponents. Implemented as gossamer.metrics.criticality.velocity_correlation + correlation_length.

Branching ratio and self-organized criticality

For cascade / avalanche processes (bundle storms in DTN, collision chains in dense swarms, role reallocation bursts in HMA) a different instrument applies. The branching ratio

\sigma = \frac{\langle n_{t+1} \rangle}{\langle n_t \rangle}

distinguishes subcritical (σ < 1), critical (σ = 1), and supercritical (σ > 1) regimes. Avalanche size distributions at σ ≈ 1 follow power laws; fitting the exponent gives the universality class directly. Implemented as gossamer.metrics.criticality.branching_ratio and avalanche_size_distribution.

Applying the toolkit

The instruments above give you a checklist for any phase-transition claim:

An order parameter with a clear direction (higher means more ordered).
Susceptibility curves across a sweep of the control parameter — does a peak appear?
Binder cumulants across multiple system sizes — do they cross?
Correlation length — does it scale with system size near the peak?
(For cascade processes) branching ratio near 1 and a power-law avalanche-size distribution.

Any paper that claims a transition without at least (1) + (2) + (3) will get sent back. The DMB + TF-ACO paper carries the full set.

References

Vicsek, T., et al. (1995). Novel type of phase transition in a system of self-driven particles. Phys. Rev. Lett. 75, 1226.
Toner, J. & Tu, Y. (1998). Flocks, herds, and schools: A quantitative theory of flocking. Phys. Rev. E 58, 4828.
Cavagna, A., et al. (2010). Scale-free correlations in starling flocks. PNAS 107, 11865.
Binder, K. (1981). Finite size scaling analysis of Ising model block distribution functions. Z. Phys. B 43, 119.