Predicting polarization fluctuations in systems of vibrated active granular particles

Abstract

Predicting fluctuations in active matter is critical to harnessing its potential. In this present study, we explore the predictability of vibrated active granular particles' (AGPs) polarization fluctuations using time series analysis. We employ an embedding-based prediction algorithm to assess how well future polarization fluctuations can be predicted from past behavior. Based on the results, predictability decreased with increasing particle density, suggesting growing system complexity. This aligns with observed “kinetic arrest” at higher densities, where AGPs exhibit rapid directional switching without sustaining a particular circulation direction. Furthermore, prediction errors increased over time, highlighting the challenge of long-term forecasting. Significant variations in predictability across individual particles also suggest potential differences in their dynamics or interactions.