Electronic structure calculation of molecules using neural network ansatz implemented with quantum Monte Carlo has witnessed an appealing advancement in the last couple of years, with contributions from several groups around the word. The method has given the community the opportunity a new tool to approach the true ground state of molecules and materials. I will discuss the latest developments to generalize the neural network wavefunction based methods to solid systems and further improvements on the accuracy and efficiency via diffusion Monte Carlo, and other architecture designs. The electronic structure calculations using neural network can also be extended to interatomic force and other quantities, which will also be discussed.  

Biosketch

Ji Chen is an assistant professor and PhD supervisor in School of Physics at Peking University. He studied physics at University of Science and Technology of China, obtained PhD degree in physics at Peking University in 2014. He subsequently did postdoc work at University College London and Max Planck Institute for Solid State Research. He was appointed assistant professor at Peking University since 2018. His main research interest is to develop cutting-edge first-principles computational methods such as quantum Monte Carlo and machine learning, and to apply the advanced computational techniques to a broad range of systems in physics, chemistry and materials science. He is an author of over 70 publications (h-index 27). More information is available at https://faculty.pku.edu.cn/chenji/en/index.htm.

Please contact phweb@ust.hk should you have questions about the talk.