Virtual Node Graph Neural Network for Full Phonon Prediction
Published in Nature Computational Science, 2024
Paper / Journal Research Briefing / Journal Front Cover / MIT News / GitHub / Data page /
Virtual node graph neural networks predict Γ-phonon spectra and full phonon dispersions directly from atomic coordinates, orders of magnitude faster than machine-learning interatomic potentials with comparable accuracy, enabling a Γ-phonon database of over 146,000 materials.
Recommended citation: Okabe, R.†, Chotrattanapituk, A.†, Boonkird, A., Andrejevic, N., Fu, X., Jaakkola, T.S., Song, Q., Nguyen, T., Drucker, N.C., Mu, S., Liao, B., Cheng, Y., & Li, M. (2024). Virtual Node Graph Neural Network for Full Phonon Prediction. Nature Computational Science 4, 522-531.
Download Paper