# CM Theory Seminar | Quantum Dynamics with Statistical Effects and Statistical Models of Quantum Effects

John Parkhill (University of Notre Dame)

The capability of electronic structure to calculate the wavefunctions, and even dynamics of large systems has improved dramatically. This has put electronic structure into an uncomfortable regime where statistical effects become as important as the correlation problem. I will discuss our efforts to describe mixed-state electronic dynamics with density matrix equations of motion, and the applications of those theories to ultrafast experiments.

Realtime mean field theories such as RT-TDDFT and RT-TDHF dominate applications because of the speed required to access picosecond timescales. Yet TDHF and TDDFT are not accurate enough to properly model resonant driving, which is only one ingredient in ultrafast spectroscopy. In this talk I discuss a simple density-matrix equation of motion implemented as an approximation to RT-TDDFT, which excites properly on resonance. Based on this foundation I compare the non-equilibrium steady states of the correct DFT and a Markovian bath model, with essentially exact results coming from HEOM showing that TDDFT can be used to study driven ultrafast dynamics. I then discuss self-consistency in correlated corrections to TDDFT which have low cost and can be applied to large systems.

Statistical sampling of molecular geometries has become an equally important issue, although empirical density functionals, which are the most practical tools for exploring geometries, make an ambiguous mixture of quantum physics and statistical modeling. I will demonstrate purely statistical models of molecular structure, and show that in the near future it is likely that purely empirical models of the PES will have several appealing advantages over empirical hybrids. of quantum mechanical models with statistics.

Realtime mean field theories such as RT-TDDFT and RT-TDHF dominate applications because of the speed required to access picosecond timescales. Yet TDHF and TDDFT are not accurate enough to properly model resonant driving, which is only one ingredient in ultrafast spectroscopy. In this talk I discuss a simple density-matrix equation of motion implemented as an approximation to RT-TDDFT, which excites properly on resonance. Based on this foundation I compare the non-equilibrium steady states of the correct DFT and a Markovian bath model, with essentially exact results coming from HEOM showing that TDDFT can be used to study driven ultrafast dynamics. I then discuss self-consistency in correlated corrections to TDDFT which have low cost and can be applied to large systems.

Statistical sampling of molecular geometries has become an equally important issue, although empirical density functionals, which are the most practical tools for exploring geometries, make an ambiguous mixture of quantum physics and statistical modeling. I will demonstrate purely statistical models of molecular structure, and show that in the near future it is likely that purely empirical models of the PES will have several appealing advantages over empirical hybrids. of quantum mechanical models with statistics.

Building: | Chemistry Dow Lab |
---|---|

Event Type: | Workshop / Seminar |

Tags: | Physics, Science |

Source: | Happening @ Michigan from Department of Physics, CM Theory Seminars |