Universal Ensemble‐Embedding Graph Neural Network for Direct Prediction of Optical Spectra from Crystal Structures (Adv. Mater. 46/2024)

Universal Ensemble-Embedding Graph-Neural Network In article number 2409175, Nguyen Tuan Hung, Mingda Li, and co-workers present a graph-neural network architecture featuring universal embedding to predict the optical properties of materials from only crystal structures. The model performs a highly precise optical prediction at the quantum simulation level, making it suitable for various applications, from solar cells to quantum materials.