Voter model with non-Poissonian interevent intervals

Area of Science:

• Computational Social Science
• Network Science
• Statistical Physics

Background:

• Human social communications often exhibit long-tail distributions in interevent intervals.
• Non-Poissonian dynamics are increasingly recognized as crucial in modeling complex systems.

Purpose of the Study:

• To investigate how non-Poissonian interevent interval distributions affect opinion formation dynamics.
• To compare consensus times in social networks under different interval distributions.

Main Methods:

• Utilized a variant of the voter model for opinion dynamics simulation.
• Numerically compared consensus times across various network structures (ring, complete graph, regular graphs).
• Analyzed the impact of exponential versus power-law interevent interval distributions.

Main Results:

• Power-law distributions of interevent intervals significantly slow down consensus formation on ring networks compared to exponential distributions.
• This slowdown is attributed to a memory effect inherent in power-law distributions.
• On complete graphs, consensus times for power-law and exponential distributions are similar.
• Regular graphs show intermediate results, with consensus slowdown decreasing as network degree increases.

Conclusions:

• The non-Poissonian nature of human interactions, specifically power-law interevent intervals, can hinder rapid opinion consensus.
• Network topology plays a critical role in mediating the effect of interevent interval distributions on opinion dynamics.
• Findings highlight the importance of incorporating realistic interaction patterns into models of social influence and collective behavior.