Features of Non-Consensus Models of Opinion Formation in Directed and Weighted Graphs

Introduction. The study concerns the dynamics of opinion formation in a non-consensus opinion model with suggestibility. This model allows for the inclusion of the agent's weighted opinion, which helps to achieve a stable state in which both opinions coexists in different proportions. Methods. We conducted numerical experiments on different modifications to the non-consensus opinion model and studied the impact of network and model parameters on the size of clusters and the stability of opinions. Results. An increase in suggestibility significantly affects the formation of clusters, decreasing their number in an equilibrium but increasing their size. Byzantine node increase the proportion of nodes with a minority opinion, but a large proportion of such nodes makes it difficult to achieve an equlibrium. In the proposed model with the community influence, only the coefficient associated with the support of the agent's opinion by his colleagues has a significant influence on the dynamics.