We consider the Hegselmann–Krause bounded confidence dynamics for n equally spaced opinions in . We completely determine the evolution when the initial separation d equals the confidence bound . Every fifth time step, three agents disconnect at either end before collapsing to a cluster. This continues until fewer than 6 agents remain in the middle, and these finally collapse to a cluster, if n is not a multiple of 6. The configuration thus freezes in time . We show that for values , the evolution is similarly periodic but with a freezing time of , and conjecture that this is maximal for equidistant configurations. Finally, we consider the dynamics for arbitrary spacings . Based on a mix of rigorous analysis and simulations, we propose hypotheses concerning the regularity of the evolution for arbitrary d, and a limiting behaviour as .
19 September 2016 in Journal of Difference Equations and Applications
We consider the Hegselmann-Krause dynamics on a one-dimensional torus and provide the first proof of convergence of this system. The proof requires only fairly minor modifications of existing methods for proving convergence in Euclidean space.