What Do Deep Neural Networks Find in Disordered Structures of Glasses?
This addresses the long-standing question of structural changes in glass transitions for soft matter physics, though it is incremental as it applies existing AI techniques to a new domain.
The study tackled the problem of identifying characteristic local structures in glasses to understand glass transitions, developing a deep neural network method that extracts these structures from particle configurations and shows they correlate with aging dynamics.
Glass transitions are widely observed in various types of soft matter systems. However, the physical mechanism of these transitions remains {elusive}, despite years of ambitious research. In particular, an important unanswered question is whether the glass transition is accompanied by a divergence of the correlation lengths of the characteristic static structures. In this study, we develop a deep-neural-network-based method that is used to extract the characteristic local meso-structures solely from instantaneous {particle} configurations without any {information} about the dynamics. We first train a neural network to classify configurations of liquids and glasses correctly. Then, we obtain the characteristic structures by quantifying the grounds for the decisions made by the network using Gradient-weighted Class Activation Mapping (Grad-CAM). We considered two qualitatively different glass-forming binary systems, and through comparisons with several established structural indicators, we demonstrate that our system can be used to identify characteristic structures that depend on the details of the systems. Moreover, the extracted structures are remarkably correlated with the nonequilibrium aging dynamics in thermal fluctuations.