My Talk at DokDok Lite 2021 on Density Functional Embedding Theory for Excited States
This talk was given at DokDok Lite 2021 (https://www.asp.uni-jena.de/doctoral-...) on September 3, 2021
I present some of my PhD work in this talk.
The work has also been published in Journal of Chemical Theory and Computation.
Read it here: https://pubs.acs.org/doi/10.1021/acs....
Abstract:
One of the main challenges in the application of accurate wavefunction theory (WFT) methods for large molecular systems is their unfavorable N^{3-8} scaling with system size. We present an implementation of a density functional theory (DFT)-based embedding framework for the calculation of excited states via WFT methods and real time-time dependent density functional theory (RT-TDDFT). The framework allows to partition the density of the total molecular system into subsystem densities: cluster (subsystem of interest) density and environment density (which is kept frozen). The influence of the environment on the cluster is manifested in the form of an embedding potential Vemb[ρclus, ρenv], which is a functional of cluster and environment densities.
The computationally intensive WFT and RT-TDDFT calculations can then be performed using only the cluster basis functions, which significantly reduces the computational time compared to a supermolecular calculation. Various strategies for constructing the embedding potential are analyzed and benchmarked with respect to excited state properties. It is found that embedding potentials containing the non-additive kinetic energy density functionals (nadd-KEDFs) offer reasonable accuracy and excellent speed-up of WFT calculations for weakly interacting systems such as solvated molecules. Furthermore, a comparison of projection operator and nadd-KEDF based embedding schemes for RT-TDDFT calculations is shown for various test systems.
All the methods are implemented within the TURBOMOLE program package [1-2].