The publication “SchNet - a deep learning architecture for molecules and materials”, by Kristof T. Schütt, Huziel E. Sauceda, Peter-Jan Kindermans, Alexandre Tkatchenko, and Klaus-Robert Müller, has been awarded with a 2024 Frontiers of Science Award. The paper introduces SchNet, a deep learning model tailored for molecules and materials, offering enhanced predictive capabilities for molecular properties and fostering insights into chemical systems through learned representations. Huziel E. Sauceda, Head of the Machine Learning for Simulations Group (ML4Sim) at the Instituto de Fisica (Mexico), and BIFOLD Alumni, accepted the honor on behalf of the research team at the International Congress of Basic Science on July 14, 2024.
The deep learning architecture SchNet is specifically designed to model atomistic systems by making use of continuous-filter convolutional layers and employs SchNet to predict potential-energy surfaces and energy-conserving force fields for molecular dynamics simulations of small molecules. The researchers demonstrated the capabilities of SchNet by accurately predicting a range of properties across chemical space for molecules and materials, where their model learned chemically plausible embeddings of atom types across the periodic table. Finally, they performed an exemplary study on the quantum-mechanical properties of C20-fullerene that would have been infeasible with regular ab initio molecular dynamics.
“This year, we focused on the application and development of machine learning algorithms in the Physical Sciences, including Materials Science, Physics, and Chemistry, achieving significant progress. In the modeling area, SchNet, a deep learning architecture, utilizes continuous-filter convolutional layers to predict quantum-mechanical interactions and potential-energy surfaces, significantly improving molecular dynamics simulations and material property predictions”, explained the jury in their decision.
With the Frontiers of Science Award the International Congress for Basic Science (ICBS) honors top research, emphasizing achievements from the past ten years which are both excellent and of outstanding scholarly value.
The award not only honors the papers' specific scientific value and originality but also acknowledges their significant impact on the research field of AI in Physics.
Publication: Schnet - a deep learning architecture for molecules and materials, K. T. Schütt; H. E. Sauceda; P.-J. Kindermans; A. Tkatchenko; K.-R. Müller
DOI: https://doi.org/10.1063/1.5019779
Presentation by H. E. Sauceda: https://www.youtube.com/watch?v=hzceG32Dm6U
Lecture Slides: https://lsp.icbs.cn/upload/1177-1719986419-Platica_ICBS_18-07-2024_red.pdf
