Mesoscopic features of a granular dynamics under cyclic compression*

We study the reversibility of rotational motion of spherical grains via molecular dynamics simulations. From the simulations, we are able to capture the motion, position, and force distribution of grains during cyclical compression and dilation and find excellent agreement with our experimental data. Additionally, we analyze the network structure formed from the contact forces between individual grains and show how the network dynamics are affected by compression amplitude as well as the evolution of the network from a transient to a steady state. Further, we use a simple machine learning approach combined with experimental data to characterize different states of our granular system.