The State of Neural Network Interatomic Potentials - IPAM at UCLA

This course aims to explore the state of neural network interatomic potentials (NNIP) and their application in atomistic simulations for research in biochemistry, material science, and reactive chemistry. The learning outcomes include understanding the strengths and weaknesses of NNIP compared to classical potentials and ab initio quantum chemistry, as well as the challenges hindering their widespread adoption. The course teaches skills such as generating training data sets, model validation, and evaluating computational speed. The teaching method involves a presentation by Justin Smith of NVIDIA, discussing NNIPs, their applications, and ongoing research efforts. The intended audience includes researchers, scientists, and professionals interested in atomistic simulation and computational chemistry.