Deep Learning Platform for Molecular Simulations


Han Wang, Institute of Applied Physics and Computational Mathematics


2018.05.11 10:00-11:00


601, Pao Yue-Kong Library


We introduce a series of deep learning based methods for molecular simulations at different scales: 1) the Deep Potential Molecular Dynamics (DeePMD): a many-body potential and interatomic forces generated by a carefully crafted deep neural network trained with ab initio data; 2) Deep Coarse-Grained Potential (DeePCG): generalization of DeePMD to the context of coarse graining; 3) Reinforced Dynamics: to use many collective variables for enhanced sampling and free energy calculation. In particular, we introduce the platform, the DeePMD-kit package, that we developed for wide applications in computational physics, chemistry, biology, and materials science.