Comparison of degree distributions of networks using adaptive rewiring at the start and end of simulations for different rewire probability λ.

Degree distributions of the network at the start (t = 0) are shown in blue and distributions at the end (t = tmax) are shown in orange. Rows correspond to different setups with respect to rewiring rules and columns correspond to different rewiring probabilities λ. The second and fourth rows show the case of disconnecting with fitness-inverse probability. If we compare them to the case of uniform disconnection (rows one and three), we see that fitness-inverse disconnection does not affect the change in degree distributions. The top two rows show reconnection with uniform probability, while the bottom two show reconnection with fitness-proportional probability. We can see that fitness-proportional disconnection results in a large number of nodes isolated from the rest of the population (equivalent to a degree d = 0). The rest of the nodes form a cluster whose degree distribution resembles that of a random network. These central clusters have marginally higher average and maximum degree compared to the networks at the start of the simulation. If we compare degree distributions along columns, we do not see any differences caused by the rewire probability λ. All simulations shown started with random networks. Results are for N = 400 and neighbor influence δ = 1. (TIF)