Deepmind published a paper in the journal Nature Physics, as well as a blog detailing the AI system and the way it can shed further light on glassy and other physical systems. As explained in the paper, the researchers determined the long-term evolution of a glass system, based on initial particle positions, using graph neural networks. In experiments, the method can predict "the locations of rearranging particles," outperforming other current methods. Modeling glassy dynamics has other uses in science, a DeepMind spokesperson told VentureBeat. In the conclusion, for example, researchers wrote that "graph networks are a versatile tool that are being applied to many other physical systems that consist of many-body interactions, in contexts including traffic, crowd simulations, and cosmology."
This is an abbreviated version of a story that first appeared in Inside AI. For the full story, click here.