Crystal graph convolutional neural networks for per-site property prediction

Graph convolutional neural networks have been shown to accurately predict materials properties by featurizing local atomic environments. However, such models have not yet been employed to predict atom-level properties such as Bader charge, magnetic moment, or site-projected band centers. In this work, we present a persite crystal graph convolutional neural network that predicts a wide array of such properties. This model captures the chemical environment around each atom and uses it to assign unique prediction labels to each site in a crystal. Using magnetic moments as a case study, we explore an example of underlying physics the per-site model is able to learn.