Transmission of kinematic information in dense granular systems: local and nonlocal network sensing

Abstract

We study how kinematic information propagates in plane strain compression tests for two granular samples, one comprising glass beads and the other is sand. Of interest are the structures of directed networks constructed from static linear relationships among grain-scale kinematical measurements obtained using digital image correlation. The exact form and kinematical information used in each linear relationship is selected using a data modelling algorithm that appeals to the information theory description length philosophy encapsulated by the principle of Occam's Razor. For both tests, we find that the observation sites with the most complicated relationships (i.e., those which require the x-..