SAPT(DFT) gone?


I was wondering what happened to the SAPT(DFT) implementation in Psi4, I saw there was a corresponding module in Psi 4.04b (beta) which disappeared in the 4.1 version. Was this module discontinued ?

SAPT(DFT) was not in previous PSI versions. However, it will be in 4.2. Do you have a particular dispersion kernel preference?

Hmm I am only aware of the adiabatic local density approximation (ALDA) exchange–correlation kernel. Are there any other good ones?

In any case I hope to use SAPT(DFT) for halogen bonded complexes (C-X…pi, C-X…N/S/O-) so any kernel which accounts well for dispersion effects would be fine. Grimme’s empirical D3 dispersion correction seems to work fine in most DFT calculations.

Hi, the ALDA kernel leads to rather large inaccuracies in SAPT(DFT). We have shown that when using a hybrid functional like PBE0 (suitably asymptotically corrected) it is far better to use a hybrid kernel with ALDA mixed from some fraction of CHF. Of course, for range-separated functionals this would need to be modified. To learn more have a look at the original papers on the theory from 2002 onwards.
Hesselmann and Jansen use only ALDA in their Molpro implementation of DFT-SAPT, but this was only possible by creating an effective local exchange potential corresponding to the non-local one from PBE0.
All the best.

1 Like