I’m looking at the examples for ALMO-EDA for bonded interactions (12.28 and 12.29) and have been wondering a few things:
- Both example molecules should have both exchange and correlation. Neither snippet uses the keyword METHOD, instead only use EXCHANGE. Why is that the case?
- Only one of the ALMO-EDA examples for non-bonded interactions specifies dispersion-free reference functionals. Should that be included in general for ALMO-EDA or not? If not, what are the criteria to judge whether to do so? And which energy contributions that are most heavily affected by their presence?
Regarding (1), using EXCHANGE for composite functionals is older Q-Chem input notation but should be backwards compatible, e.g., METHOD = B3LYP should be equivalent to EXCHANGE = B3LYP (w/o specifying CORRELATION in the latter case). Or in the examples in question, EXCHANGE = wB97X-D is equivalent to METHOD = wB97x-D, while EXCHANGE = HF (w/o CORRELATION) is equivalent to Hartree-Fock. You should see the functional printed out just before the SCF iterations begin, so you can confirm.
I’ll leave (2) for someone else.
Thanks for the clarification!
The dispersion-free functional is set to Hartree-Fock exchange by default (no correlation). This seems to work well for dispersion-corrected hybrid functionals such as wB97M-V, B3LYP-D3(BJ), etc. For dispersion-corrected semi-local functionals like B97M-V this can still be used, whereas another recommended option is the revPBE functional (DISP_FREE_X = revPBE, DISP_FREE_C = PBE).
The choice of this dispersion-free functional would only affect the partition between Pauli repulsion and dispersion as components of the frozen interaction energy.