NBO analysis with different methods


I ran NBO analysis on a set of ~ 100 small molecules (4 heavy atoms max), and compared the NBO analysis with HF and B3LYP (both using the 6-311+G* basis set).

In my results, I find that NBO hyperconjugation energies obtained from the NBO analysis with the DFT wavefunction/density are basically the same (but scaled down) as those obtained with the HF wavefunction given as input.

Is that a normal behaviour to observe? Shouldn’t the wavefunctoin/density matrix be rather different when comparing HF and B3LYP? Has anyone else noticed this in the past?

Thanks !

What version of Psi4 are you using?


I actually ran these calculations with GAMESS, as Psi4 does not suppoer (yet?) NBO analysis.

I just hoped I could ask here.


Psi4 does support NBO analysis, although I’m aware other programs support some features that Psi does not yet. (For instance, the $DEL keyword.)

There were some bugs in Psi4’s NBO support. I fixed one of them for Psi4 1.3.2, and the other bug is fixed in the developer version and will be available in Psi4 1.4.

You can see an example of NBO analysis in Psi using the 1.4 interface here.

As for expected differences between HF and B3LYP for hyperconjugation energies, I don’t know.