Colloquia

In this talk, I will present our recent efforts on dynamical simulations of correlated electron systems. I will first discuss new quantum molecular dynamics (QMD) methods based on advanced many-body techniques, such as Gutzwiller/slave-boson and dynamical mean-field theory, that are capable of modeling strong electron correlation phenomena. We apply our new QMD to simulate the correlation-induced Mott transition in a metallic liquid, and the nucleation-and-growth of Mott droplets in Hubbard-type models. I will also discuss the implementation of the ab initio Gutzwiller MD for simulating hydrogen liquids under high pressure. To overcome the obstacle of huge computational complexity in such large-scale simulations, I will discuss how simulation efficiency can be significantly improved with the aid of modern machine learning methods. In particular, deep-learning neural-network holds the potential of achieving large-scale quantum-accuracy simulation of correlated systems without the electrons.