Opinion dynamics in multi-agent systems under proportional updating and any-to-any influence (Q2216012)

This paper studies an agent-based opinion dynamics for all to all communication. A group of \(n\) agents belong to \(c\) classes. The number of agents in class \(i\) is \(n_i\) with \(n=\sum_{i=1}^cn_i\). If the agent \(j\) is in the class \(i\), its opinion is denoted by \(v_{ij}\in[-1,1]\). The coefficient \(\gamma_{kl}\in[0,1]\) describes the intensity of interaction between classes with \(1\le k,l\le c\). The system is given as \(v_{ij}(t+1)=v_{ij}(t)-\sum_{k=1,k\not=i}^c\gamma_{ik}\sum_{l=1}^{n_k}(v_{ij}(t)-v_{kl}(t)) +\gamma_{ii}\sum_{m=1,m\not=j}^{n_i}(v_{im}(t)-v_{ij}(t))\). By imposing some conditions on the coefficient \(\gamma_{kl}\), the authors obtain the closure property for this dynamical system. The paper shows that the opinion of each agent can converge in the long term for this model and provide the steady-state solution of the multiagent system. For the entire collection see [Zbl 1452.90001].