I find that in fchk files from MP2 calculations on a couple of small molecules that the total densities appear to be mislabelled. The density labelled “Total SCF Density” is different from the “Total SCF Density” from a Hartree-Fock calculation, so it is presumably the “Total MP2 Density”. It is however identical to another density that is labelled the “Total CC Density”, though it wasn’t a CC calculation. It’s a detail, but it would be good to have the labelling clarified.
History of the
"Total CC Density" is in this comment block. I agree that labeling isn’t the best, but as an unregulated intermediate file, we didn’t want to move too far from fchk “standards”. Is the second line of the fchk with type,method,basis any more informative?
The “Total SCF Density” != “Total SCF Density” from HF sounds more concerning. Definitely running
OK, I understand the use of “Total CC Density” though I find it somewhat unsatisfactory.
As for “Definitely running
gradient('mp2', ...) ?” I may have missed something. I used
energy, wfn = energy('mp2', return_wfn=True). Does this not return the MP2 wavefunction? If it doesn’t, why do I see a “Total CC density” section in the fchk file? I don’t see anything in the Psi4 manual to clarify this.
It does return a MP2 “wavefunction”, but that contains what’s been computed thus far, which for
energy('mp2' is MP2 energy and HF orbitals. There’s an example right above http://psicode.org/psi4manual/master/gdma.html#options that may be helpful.
Ah, thanks. I hadn’t looked at the gdma documentation because I’m using my own gdma code. But it’s questionable whether Psi4 should generate an fchk file containing something labelled as the MP2 (or CC) density when it is actually the SCF density.
IIRC there was a bug in Psi4 that ref_wfn returns the correlated density, not the un-correlated density. #1475 fixed the checkpoint file writer, and did not touch the machinery of Wavefunction which is a separate issue. It’s quite possible that this bug still exists and should be fixed as well.