Dresden 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

TT: Fachverband Tiefe Temperaturen

TT 23: Correlated Electrons – Poster I

TT 23.27: Poster

Monday, March 9, 2026, 18:00–20:00, P1

Stochastic semiclassical electron-lattice dynamics with DMFT — •Tom Kahana, Francesco Valiera, and Martin Eckstein — I. Institute of Theoretical Physics, University of Hamburg, Hamburg, Germany

The interplay between lattice motion and strongly correlated electronic phases poses a major challenge in nonequilibrium condensed-matter systems, especially near phase transitions where ionic distortions strongly affect the electronic state. We address this problem by solving the coupled electron-lattice dynamics using a quasi-equilibrium (adiabatic) Dynamical Mean-Field Theory on an inhomogeneous lattice defined by the instantaneous phonon configuration. Conservative forces, friction, and stochastic contributions to the lattice dynamics are obtained from the electronic correlation functions. Because each lattice site must be treated as an impurity embedded in a self-consistent bath, this approach is computationally demanding. To make it feasible, we replace the DMFT impurity solver with a trained neural-network emulator that reproduces high-order impurity solutions. This NN-accelerated DMFT loop enables computationally efficient simulations of the coupled electron-lattice dynamics on inhomogeneous lattices. We present benchmarks for the Hubbard-Holstein model. The method provides a path toward simulations of photo-induced transitions in correlated materials, such as Mott and charge-ordered states.

Keywords: Dynamical Mean-Field Theory; Neural Networks; Electron-lattice dynamics; Strongly correlated electrons; Hubbard-Holstein model