【凝聚态物理-北京大学论坛 2024年第13期（总595期）】Applications of artificial intelligence in materials and chemistry

发布日期:2024-05-07 点击数:

主讲人:	Prof. Pavlo O. Dral (Xiamen University)
地点:	物理大楼西563会议室
时间:	2023年5月30日（星期四）下午3:00－4:30
主持联系人:	陈基 ji.chen@pku.edu.cn
主讲人简介:	Pavlo O. Dral is a Full Professor at Xiamen University and an Assistant Dean in international admissions matters at Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University. His research is focused on accelerating and improving quantum chemistry with artificial intelligence/machine learning. Pavlo Dral is a founder of the MLatom package for atomistic machine learning simulations and a co-founder of the Xiamen Atomistic Computing Suite. In 2021, he was awarded an Outstanding Youth (Overseas) by the National Natural Science Foundation of China. Pavlo Dral did his PhD with Prof. Tim Clark at the University of Erlangen–Nuremberg in 2010–2013 and a postdoc with Prof. Walter Thiel at the Max Planck Institute for Coal Research in 2013–2019. He is the gold medal winner of the 36th International Chemistry Olympiad, 2004 in Kiel, Germany. Pavlo O. Dral is an Editor of two open-access journals Artificial Intelligence Chemistry and SciPost Chemistry. He is the author of over 70 publications cited 5000 times (h-index 27). More information is available on Dral’s group website dr-dral.com.

I will present our work on developing AI methods and their application for performing physical chemistry simulations accurately and fast. This work is mostly based on our MLatom@XACS platform[1] that seamlessly integrates various machine learning (ML) and quantum chemical (QC) methods and their combinations for a wide variety of simulations. A non-exhaustive list of MLatom’s capabilities includes ab initio, DFT, semi-empirical QC methods, various machine learning potentials (equivariant MACE, popular ANI and DeepPot-SE, etc.), and their combinations such as universal AIQM1[2] and ANI-1ccx ML-based methods. MLatom supports thermochemical calculations, molecular and quantum dynamics[3], spectra simulations,[4] and more. We also implemented efficient active learning procedures based on MLatom. Many of the simulations can be run online at the https://XACScloud.com.

Representative works:

[1] P. O. Dral, F. Ge, et al. MLatom 3: A Platform for Machine Learning-Enhanced Computational Chemistry Simulations and Workflows. J. Chem. Theory Comput. 2024, 20, 1193–1213.

[2] P. Zheng, R. Zubatyuk, W. Wu, O. Isayev, P. O. Dral. Artificial Intelligence-Enhanced Quantum Chemical Method with Broad Applicability. Nat. Commun. 2021, 12, 7022.

[3] A. Ullah, P. O. Dral. Predicting the future of excitation energy transfer in light-harvesting complex with artificial intelligence-based quantum dynamics. Nat. Commun. 2022, 13, 1930.

[4] P. O. Dral, M. Barbatti. Molecular Excited States Through a Machine Learning Lens. Nat. Rev. Chem. 2021, 5, 388–405.