Tsinghua Shenzhen International Graduate School
Professor Kuang Yu graduated from Peking University in 2008 and obtained his PhD in theoretical chemistry in UW, Madison in 2013, supervised by Professor J. R. Schmidt. Then he worked as a postdoc in Professor Emily Carter’s group in Princeton University, from 2013 to 2016. He later joint D. E. Shaw Research as a research scientist in 2016, until he became an Assistant Professor in Tsinghua-Berkeley Shenzhen Institute (later merged into Tsinghua Shenzhen International Graduate School) in 2018.
Accurate simulation of molecular systems has important applications in fields such as materials and drug design. However, the fidelity of molecular force fields limits the accuracy of molecular dynamics (MD) simulations, and reliable force field parameterization has always been a bottleneck for high-throughput screening. This report will focus on how to introduce modern artificial intelligence (AI) technologies (e.g., neural network models and auto differentiation) into force field development. The report is divided into two parts: bottom-up and top-down. From the bottom-up, employing both short-/long-range and bonding/nonbonding separations, we combine the flexibility of neural networks and the robustness of physical models. Bulk force fields with chemical accuracy can be constructed using only tiny cluster data, demonstrating great transferability in molecules of different sizes and physical environments. On the other hand, from the top-down, we have developed a high-throughput automated force field optimization platform (named DMFF[1]) based on auto differentiation technique. It realizes automated parameter tuning based on gradient descent algorithms, greatly improving the efficiency, the throughput, and the scalability of traditional force field optimization. Through different cases, we will demonstrate the important applications of modern AI technology in the development of molecular models in different scales.
©khalifa university. All rights reserved.
Funded by
