Multi-Scale Dynamical Modeling of Correlated Systems Based on Dynamical Mean-Field Theory

In this thesis, we introduce a novel scheme of adiabatic quantum molecular dynamics (QMD), in which the electron degrees of freedom are integrated out on the fly by the dynamical mean-field theory (DMFT) calculation. Compared with the QMD based on the popular density functional theory, our new scheme is able to describe phenomena due to strong electron correlation, such as Mott metal-insulator transition (MIT). Moreover, our DMFT-QMD also provides information on the incoherent non-quasi-particle electronic excitations, thus significantly generalizing the capability of Gutzwiller/Slave-boson-based QMD recently developed by our group. We use this new MD method to study the Mott transition in an atomic liquid of hydrogen-like atoms. We observe a reentrant phase transition driven by Coulomb repulsion and obtain various nontrivial atomic and electronic properties of the system.

Additionally, we combine exact diagonalization with molecular dynamics simulation to study the correlated liquid model in the strong coupling limit, in which the local moments on atoms dominate. We discover a tendency of dimer formation in the system and the dimers undergo a dissociation process driven by Coulomb repulsion.

Our work opens a new avenue for multi-scale dynamical simulations and modeling of strongly correlated electron systems. Our results can provide unique insights into the dynamics and the electronic properties of the MIT in correlated liquid systems.