The Bass diffusion model: agent-based implementation on arbitrary networks

Abstract

We show how the combined use of the free software packages networkX and NetLogo allows to implement quickly and with large flexibility agent-based network simulations of the classical Bass diffusion model and of its extensions and modifications. In addition to the standard internal graph implementations available in NetLogo (random, Barabasi-Albert-1 and small world), one can thus employ more complex Barabasi-Albert and small-world networks, plus scale-free networks with arbitrary power-law exponent $\gamma$ built in networkX through a configuration model algorithm. It is also possible to induce degree correlations in the networks in a controlled way via Newman rewiring and to simulate dynamics on arbitrary signed networks (networks where link can have positive or negative weights, with corresponding effects on diffusion). Some new results obtained in the agent-based simulations (and differing from those in mean-field approximation) are the following. The introduction of assortative correlations in scale-free networks has the effect of delaying the adoption peak in the Bass model, compared to the uncorrelated case. The peak time $t_{max}$ depends strongly also on the maximum degree effectively present in the network. For diffusion models with threshold on signed network, if negative influences have a weight equal to or greater than positive influences, then a high level of clustering tends to cause adoption blockades. In connection to this, by analysing statistical ensembles of assortative scale-free networks generated via Newman rewiring one observes a remarkable strong correlation between the function of the average degree of first neighbors $\bar{k}_{nn}(k)$ and the average clustering coefficient depending on the degree $\bar{C}(k)$.