Figure 6. Prediction quality for the limit 1/κm of sustained activity (transition point B) when increasing the edge density.
The data are obtained by scanning graphs having a fixed number N = 80 of nodes and a varying number M = 150...2000 of edges (steps of 30), considering 50 realizations for each value of M, and for each network all the possibilities for the input node. The plot displays the quality of the predictions 1/κm = kmax (maximal degree, blue dashed curve) and 1/κm = kmax,1 (maximal degree in the first layer, red full curve) compared to the numerical value of 1/κm (rounded to the nearest integer). Each point in these curves is the number of successful predictions over 4000 trials, normalized so that a perfect prediction would get a score of 100. Upper panel: ER graphs; lower panel: BA graphs.