Computed thermodynamic energies from DF-MP2 freq calculation

Hello all,

First time posting here. I have been using Psi4 this past year, and have really enjoyed it - and thanks to the developers for their hard work.

I noticed that when I compute frequencies with DF-MP2 (i.e. frequency(‘mp2’)), the reported thermodynamic energy totals use the SCF energy. Is there a reason why they don’t use the DF-MP2 energy? I was doing this by hand, to ensure that I understood what Psi4 was doing and reporting, when I came across this.

I also cannot reproduce the SCF-based reported total energies exactly. By using the final optimized SCF energy (via an optimize(‘mp2’) calculation), I compute total thermodynamic energies that are off by ~0.0005 Hartrees (~0.03 kcal/mol).

Below is the input and selected output of the results for ethane. I also include the Reference Energy (i.e. SCF) and Total Energy (i.e. DF-MP2) obtained from the original optimization run.

I am using Psi4 0.4.183 Driver / Git: Rev {master} 5801071 version.

Thanks in advance,

memory 30 Gb

molecule {
C 0.762214375704 0.000000349718 -0.000000563472
H 1.156979457961 0.499974894010 -0.887950178060
H 1.156985286530 0.519001924519 0.876963063417
H 1.156986023525 -1.018973223609 0.010981812258
C -0.762214375704 -0.000000349718 0.000000563472
H -1.156979457961 -0.499974894010 0.887950178060
H -1.156985286530 -0.519001924519 -0.876963063417
H -1.156986023525 1.018973223609 -0.010981812258

set globals {
basis 6-31G(d)

final DF-MP2 energies from the optimization calc.:
Reference Energy = -79.2285468482043882 [H]
Total Energy = -79.5037603652802431 [H]

selected output from the frequency calc.:
Energies in Hartree/particle: Correction Total
Energy (0 K) 0.07723220 -79.15131465
Internal energy 0.08065684 -79.14789001
Enthalpy 0.08160103 -79.14694582
Gibbs Free Energy 0.05415999 -79.17438686

maybe try the latest version? V 1.0 was just released.

Thanks hokru - I reran the calculations using Psi4 1.0 and now the corrected thermodynamic energies use the DF-MP2 electronic energy. Problem solved.