Apologies if this seems like it is adequately covered in the documentation. I'm new both computational chemistry and NWChem, so I may just not have enough background knowledge to fully understand the documentation.
We're just starting a new project where we are hoping to calculated the energies of the excited states of a number of molecules using TDDFT. We've been successful in getting NWChem to run and give us output, and even are able to view the output, however we're looking at comparing TDDFT calculations using the Tamm Dancoff approximation. The documentation for NW Chem implies that it has options for CIS, RPA a.k.a. TDHF, TDDFT and TDA as separate options when I read this in the overview:
NWChem supports a spectrum of single excitation theories for vertical excitation energy calculations, namely,
configuration interaction singles (CIS), time-dependent Hartree-Fock (TDHF or also known as random-phase
approximation RPA), time- dependent density functional theory (TDDFT),[ref] and Tamm-Dancoff approximation
to TDDFT. These methods are implemented in a single framework that invokes Davidson's trial vector algorithm
(or its modification for a non-Hermitian eigenvalue problem).
And I've recently come across a paper, which doesn't use NWChem, and also implies that CIS, RPA, TDDFT and TDA are four separate analytical methods for calculating excited states. See here <http://pubs.acs.org/doi/abs/10.1021/ct400597f> In addition to saying that CIS and RPA are not as successful as TDDFT and TDA.
However
Further in the NWChem documentation for the TDDFT module it suggests that my choices are two: RPA (the default) and CIS.
We have tried running both ways and can certainly attest to the fact that CIS is computationally faster, but we get different results between the two methods and are trying to determine the best way for us to go forward (is faster also better?)
Am I missing something? Is RPA the same as TDDFT and CIS the same as TDA?
|