Swarmalators
Swarmalators are generalizations of phase oscillators that swarm around in space as they synchronize in time.
≈ 5 min read
Names in other languages
No names in other languages are held for this entry yet.
Article bodies are English and are not machine-translated. A name in another language is not a translation of the article.
Swarmalators are generalizations of phase oscillators that swarm around in space as they synchronize in time. They were introduced to model the diverse real-world systems which both sync and swarm, such as vinegar eels, magnetic domain walls, and Japanese tree frogs.
More formally, they are dynamical units with spatial degrees of freedom and internal degrees of freedom whose dynamics are coupled. \dot{x}_i & = & \frac{1}{N} \sum_j I_{att}(x_j-x_i) F(\theta_j - \theta_i) - I_{rep}(x_j - x_i) \\ \dot{\theta}_i & = & \frac{1}{N} \sum_j G_{sync}(\theta_j - \theta_i) H(x_j - x_i) \end{array} The spatial dynamics combine pairwise interaction I_{att} with pairwise I_{rep} , which produces swarming / aggregation. The novelty is the attraction is modified by a phase term F ; thus the aggregation becomes phase-dependent.
Likewise, the phase dynamics contain a sync term G_{sync} modified by a spatial term so the synchronization becomes position dependent. In short, the swarmalators model the interaction between self-synchronization and self-assembly in space.
While in general the position could be in 2D or 3D, the instance of the swarmalator model originally introduced is a 2D model x \in \mathbb{R}^2 and the choices for I_{att}(x), I_{rep}(x) etc. were \begin{array}{rcl} \dot{x}_i & = & \frac{1}{N} \sum_j \frac{(x_j - x_i)} \left(1 + J \cos(\theta_j - \theta_i)\right) - \frac{x_j - x_i} \end{array} There are two parameters J and K : J controls the strength of phase-space attraction/repulsion, while K describes the phase coupling strength. The above can be considered a blending of the aggregation model introduced from biological swarming (the spatial part) and the Kuramoto model of phase oscillators (the phase part).
W_{\pm} & = & S_{\pm} e^{i \phi_{\pm}} = &\frac{1}{N} \sum_j e^{i(\phi_j - \theta_i)} \end{array} where \phi_i := \arctan(y_i / x_i) are used. Figure 2 plots S_{\pm} versus K for fixed J . As can be seen, S_+ = 1 in the rainbow-like static phased wave state (at K = 0), and then declines as K decreases.
A second order parameter \gamma , defined as the fraction of swarmalators that have completed at least one cycle in space and phase after transients in also plotted, which can distinguish between the active phase wave and splintered phase wave states. \dot{x}_i & = \nu_i' + \frac{J'}{N} \sum_j \sin(x_j - x_i) \cos(\theta_j - \theta_i) \\ \dot{\theta}_i & = \omega_i' + \frac{K'}{N} \sum_j \sin(\theta_j - \theta_i) \cos(x_j - x_i) \end{array} where \nu_i, \omega_i are the (random) natural frequencies of the i-th swarmalator and are drawn from certain distributions g(.) . This 1D model corresponds to the angular component of the 2D swarmalator model.
The restriction to this simpler topology allows for a greater analysis. For instance, the model with natural frequencies can be solved by defining the sum/difference coordinates \xi_i, \eta_i = x_i \pm \theta_i the model simplifies into a pair of linearly coupled Kuramoto models \begin{array}{rcl} \dot{\xi}_i &= \nu_{i} + K r \sin(\phi - \xi) + J s \sin( \psi - \eta) \\ \dot{\eta}_i &= \omega_{i} + J r \sin(\phi - \xi) + K s \sin( \psi - \eta) \end{array} where K,J = (J' \pm K')/2 and \nu, \omega = \nu' \pm \omega' and the rainbow order parameters are the equivalent of the 2D model \begin{array}{rcl} Z_{\xi} = r e^{i \phi } := N^{-1} \sum_j e^{i (x_j + \theta_j)} \\ Z_{\eta} = s e^{i \psi} := N^{-1} \sum_j e^{i (x_j - \theta_j)} \\ \end{array} For unimodal distribution of \omega, \nu such as the Cauchy distribution, the model exhibits four collective states depicted in the figure on the right.
Async or (r, s) = (0,0) state. Swarmalators do not exhibit any coherence either in space or phase, being distributed uniformly in position and phase. This state is characterized by the absence of any synchronization or spatial clustering among the swarmalators, as reflected by the zero values of both order parameters r = s = 0 .
Phase wave or (r, s) = (S_{pw},0) state. In the phase wave state, swarmalators form a band or wave pattern with the position x_i and phase \theta_i are correlated. The wave can either run clockwise or counterclockwise.
Mixed or (r, s) = (S_1,S_2) state. Swaramlators again form a phase wave, but now the wave is distorted, forming two rough clusters; thus it is the a mixture of the phase wave and the sync state (described next). Sync or (r, s) = (S_{sync},S_{sync}) state.
Swarmalators form two synchronous cluster in both space and phase. Single cluster states are formed for some initial conditions. Note in each state, the swarmalators split into a locked/drifting sub-populations, just like the Kuramoto model.
The locked population are the denser regions in the Figure, the drifters the light grey regions. The figure to the right compares the bifurcations of the Kuramoto model to those of the 1D swarmalator model.
For the Kuramoto model (top row), the sync order parameter R := |\langle e^{i \theta} \rangle| bifurcates from the async state ( R = 0 ) and then increases monontonically in the sync state ( R > 0 ). For the 1D swarmalator model, the bifurcations are richer. Sweeping K from negative to positive values triggers the sequence of bifurcations from the async to the sync state.
Text derived from https://en.wikipedia.org/w/index.php?oldid=1368852804 under CC BY-SA 4.0; modifications have been made.