epl draft Opinion Formation on Multiplex Scale-free Networks
Vu X Nguyen,Gaoxi Xiao,Xin-Jian Xu,Guoqi Li,Zhen Wang
Abstract:Most individuals, if not all, live in various social networks. The formation of opinion systems is an outcome of social interactions and information propagation occurring in such networks. We study the opinion formation with a new rule of pair-wise interactions in the novel version of the well-known Deffuant model on multiplex networks composed of two layers, each of which is a scale-free network. It is found that in a duplex network composed of two identical layers, the presence of the multiplexity helps either diminish or enhance opinion diversity depending on the relative magnitudes of tolerance ranges characterizing the degree of openness/tolerance on both layers: there is a steady separation between different regions of tolerance range values on two network layers where multiplexity plays two different roles respectively. Additionally, the two critical tolerance ranges follow a one-sum rule; that is, each of the layers reaches a complete consensus only if the sum of the tolerance ranges on the two layers is greater than a constant approximately equaling 1, a double of the critical bound on a corresponding isolated network. A further investigation of the coupling between constituent layers quantified by a link overlap parameter reveals that as the layers are loosely coupled, the two opinion systems co-evolve independently, but when the inter-layer coupling is sufficiently strong, a monotonic behavior is observed: an increase in the tolerance range of a layer causes a decline in the opinion diversity on the other layer regardless of the magnitudes of tolerance ranges associated with the layers in question. Introduction. – Opinion formation is an interesting topic in the studies on complex networks [1, 2]. It helps to predict opinion distribution in a population after a temporal process of spreading through social interactions [3, 4]. Such studies may reveal that local pair-wise interactions of individuals eventually form a global equilibrium. Numerous opinion spreading models have been proposed to reveal the properties of opinion systems by introducing various rules of communication between individuals. Some of the most well-known models include Voter model [5, 6], major rule model [7], Sznajd model [2], etc. Proposed by Deffuant et al. in 2000, the Deffuant model [8], as one of the most popular bounded confidence models, has grabbed much attention over the past decades. The model employs the parameter of tolerance range. (a)egxxiao@ntu.edu.sg (b)zhenwang0@gmail.com Such a parameter to a certain extent reflects the concept of selective exposure, which describes one’s tendency to involve himself/herself in social interactions that support his/her opinions [9, 10, 11]. It is found that the tolerance bound (also interpreted as confidence bound, tolerance range, or uncertainty threshold [12, 13, 14]) plays an important role in opinion formation processes. The numerical experiments reveal an interesting result, that is, the number of coexisting opinion clusters in the steady state is approximately 1/(2 ∗ tolerance range) [8, 15]. A notion of critical confidence bound is also captured to refer to a threshold beyond which an opinion system eventually converges to a single cluster sharing the same opinion that equals the average of first impression [8, 16, 17]. Some analytical frameworks have been developed using the mean-field approach [18, 19, 20].