Reference: Martens E. A., Klemm, K., Transitions from Trees to Cycles in Adaptive Flow Networks (under review), or arXiv:1711.00401 for a pre-print.
Simulations of the adaptive dynamics for networks with height H = 4. The flow at the top of the tree-like structure (inlet) is constant, while the outlets in the leaves experience load fluctuations of strength a. During the simulation, the parameter is quasi-adiabatiacally decreasing from 1.0 in steps of size 0.01.
Legend for edge thickness and conductances Ckl:
Supplementary Video 1: Augmented tree with cross-edges in minimal triangular subgraphs
Dynamics for a tree augmented with cross-edges allowed only within triangular minimal subgraphs (Fig. 1 c)).
Supplementary Video 2: Augmented tree with all cross-edges.
Dynamics for a tree augmented with all possible cross-edges allowed (Fig. 1 d))
Supplementary Video 3: Damaged network, augmented tree with all cross-edges, realization #1.
Damaged versions of the network in Video 2, but where roughly half of all cross edges have been randomly selected and removed.
These simulations confirm the observation that the cross edges begin to conduct in order of the tree hierarchy, starting from sinks and moving towards the source.
Supplementary Video 4: Damaged network, augmented tree with all cross-edges, realization #2.
Another instance of a damaged network in Video 2, but where roughly half of all cross edges have been randomly selected and removed.
These simulations confirm the observation that the cross edges begin to conduct in order of the tree hierarchy, starting from sinks and moving towards the source.
Reference: Martens E. A., Panaggio, M., Abrams, D.M. Basins of Attraction for Chimera States, New Journal of Physics, 2016 arXiv:1507.01457 for a pre-print.
Destination maps as a function of 0.1≤s≤1 (A=0.2,β=0.025). Even though A=0.2 is fairly large with regards to our perturbative calculus, numerical results match the predicted motion qualitatively well. As s increases from zero, basins merge and pinch-off in an alternating fashion, so that the basin boundaries rotate counter-clockwise about R0 (d,ψ)=(0,0). Once s reaches sc≈√(1-A), this rotation stops, demonstrating that knowledge of the trajectory position in the s=sc plane is sufficient for determining the final fate of the trajectory.
Basins are color-coded, where blue/red/yellow denote the DS/SD/SS0 basins, respectively, where S and D denote which one of the populations is synchronized.
Supplementary Video 2: Double helical structure of trajectories near R0-manifold.
Twisting motion of trajectories in a double helical structure following the R0-manifold (A=0.1,β=0.025).
Initial conditions of 31 trajectories are equally spaced with s=0.1045; -0.0345≤d≤0.0345; ψ=0.
Trajectories are color-coded, depending on which state they converge to (blue/red/yellow denote the DS/SD/SS0 basins, respectively, where S and D denote which one of the populations is synchronized).
Supplementary Video 3: Double helical structure of trajectories near Rπ-manifold.
Twisting motion of trajectories in a double helical structure following the Rπ manifold (A=0.1,β=0.025). Initial conditions of the 3 trajectories are s=0.4487,d ∈ {-0.6,-0.2,0.6}*10-3,ψ=π.
Supplementary Video 4: 3D-visualization of separatrices.
Three dimensional visualization of the separatrices emanating from the chimera saddle points near the R0-manifold
(A=0.1, β=0.025).