Fig. 1: Graph deep learning architecture for materials science.

V_n and E_n denotes the set of node/atom ({v_i}) and edge/bond features ({e_ij}), respectively, in the n^th layer. Some implementations include a global state feature (U) for greater expressive power. Between layers, a sequence of edge (f_E), node (f_V) and state (f_U) update operations are performed. f_E, f_V and f_U are usually modeled using multilayer perceptrons. In the final step, the edges, nodes and state features are pooled (P) and passed through a multilayer perceptron to arrive at a prediction.