Sznajd Complex Networks

TL;DR

This paper introduces a novel geographic complex network model based on a feedbacked Sznajd dynamics, analyzing its properties and community structures compared to traditional models.

Contribution

It presents a new network growth scheme using Sznajd dynamics on edges, revealing unique community and hierarchical properties.

Findings

Networks exhibit patches of connected communities.

Distinct hierarchical measurement profiles compared to random and scale-free networks.

Identifies properties related to feedback-driven network topology.

Abstract

The Sznajd cellular automata corresponds to one of the simplest and yet most interesting models of complex systems. While the traditional two-dimensional Sznajd model tends to a consensus state (pro or cons), the assignment of the contrary to the dominant opinion to some of its cells during the system evolution is known to provide stabilizing feedback implying the overall system state to oscillate around null magnetization. The current article presents a novel type of geographic complex network model whose connections follow an associated feedbacked Sznajd model, i.e. the Sznajd dynamics is run over the network edges. Only connections not exceeding a maximum Euclidean distance $D$ are considered, and any two nodes within such a distance are randomly selected and, in case they are connected, all network nodes which are no further than $D$ are connected to them. In case they are not connected, all nodes within that distance are disconnected from them. Pairs of nodes are then randomly selected and assigned to the contrary of the dominant connectivity. The topology of the complex networks obtained by such a simple growth scheme, which are typically characterized by patches of connected communities, is analyzed both at global and individual levels in terms of a set of hierarchical measurements introduced recently. A series of interesting properties are identified and discussed comparatively to random and scale-free models with the same number of nodes and similar connectivity.