Open venkatachalapathy opened 5 years ago
I'm going to claim that 0 ≤ λ ≤ 4 is plenty enough range, based on the plots we're seeing.
Let us focus only on increasing the number of vertices.
Actually, I changed my mind. I want to make sure that the Gini is actually plateauing out.
I want to make sure that the Gini is actually plateauing out. By plotting a larger (higher) range of λ values? Or are you saying more trials per value to narrow error bars, or...?
Oh. higher range of $\lambda$ values. (see my #analysis channel comments from yesterday night for more details). Also, this is going to help us connect with all-to-all networks.
Building on #62, we can increase the no of nodes/agents in the network to look for sharper transitions in both no of components and in Gini as we cross the $\lambda=1$ threshold. Also, as discussed in a mini-zoom couple of days back, the parameter sweep should go beyond $\lambda=6$ just to make sure we see the trend.