Fig. 3: Benchmarking neural-network non-adiabatic molecular dynamics (N2AMD) on nonadiabatic molecular dynamics (NAMD) simulation. | Nature Communications

Fig. 3: Benchmarking neural-network non-adiabatic molecular dynamics (N2AMD) on nonadiabatic molecular dynamics (NAMD) simulation.

From: Advancing nonadiabatic molecular dynamics simulations in solids with E(3) equivariant deep neural hamiltonians

Fig. 3

a, b Absolute values of nonadiabatic couplings (NACs) between valence band maximum (VBM) and conduction band minimum (CBM) along a short trajectory for (a) TiO2 and (b) GaAs, calculated by N2AMD and density functional theory with hybrid functional (DFT-HSE06) respectively. c, d The demonstrative non-radiative electron-hole recombination for TiO2 and GaAs, calculated by N2AMD and DFT-HSE06 respectively. (Inset of c and d. Geometry structure of TiO2 and GaAs.) e Comparison of computational resources for one single-step calculation in NAMD required by two methods. Source data are provided as a Source Data file.

Back to article page