Figure 3: Topological properties of NNG equilibrium networks as a function of the power-law exponent.
From: Navigable networks as Nash equilibria of navigation games
Panel (a) shows the total cost (number of edges), average clustering 
and stretch in NNG-simulated networks as functions of γ. Stretch (shown in the inset) is the average factor showing by how much longer the greedy navigation paths are, compared with the shortest paths in the network. Stretch equal to 1 means that all navigation paths are shortest possible. The plotted points are mean values while the error bars show minimum and maximum values obtained for the NNG over 10 random sprinkling of nodes for a given value of γ. Panel (b) shows the success ratio as a function of the percentage of edges randomly deleted from the network. The smaller the γ, the more robust the navigability with respect to this network damage.
