Efficient methods for ab-initio non-adiabatic molecular dynamics

Abstract:
The aim of this project is to develop, assess, and apply efficient methods for ab-initio non-adiabatic molecular dynamics. In particular, we intend to circumvent the explicit calculation of the non-adiabatic couplings, which typically represents a computational bottleneck. As has been shown by Persico et al., the probabilities of non-adiabatic transitions can be approximately calculated by a local diabatization procedure and we would like to transfer this approach (so far implemented within semiempirical MD) to ab-initio MD based on multireference CI and on time-dependent DFT techniques. The developed methods will be applied to photochemistry of organic and biologically relevant compounds (in collaboration with H. Lischka) and to study of time-dependent optical properties and reactivity of small metal clusters and their complexes with organic ligands (in collaboration with V. Bonacic-Koutecky).