ORCA parser does not include dispersion correction in scfenergies

Hi,

Currently, the ORCA parser does not take into account dispersion corrections provided through post-SCF corrections, like the Grimme D3 correction.

I have attached an example below.

wB97XD3_def2-SVP_toluene_S0_TDA6S_2-20.txt

In this example, the pre-D3 SCF energy on the Total Energy : line is -306.98073026 Eh, which is what cclib finds. However, the DFT+D3 energy is on the FINAL SINGLE POINT ENERGY line and is -306.983847616167 Eh.

I appreciate that this is possibly ‘intended’ behaviour, since strictly speaking, the D3 correction is post-SCF. However, this is inconsistent with, for example, Gaussian, since Gaussian includes the dispersion component in print outs of the SCF energy.

[langner]

It’s great we’re getting around to fixing this. It sounds like my intuition is in the minority and most people expect scfenergies to include any dispersion corrections, and I’m OK with that as well. If everyone agrees, let’s go ahead and bake that into the docs and merge the ORCA fix.

[ghutchis]

I’d like to chime in. Whatever people do, needs to be broadly advertised. I, for one, assumed that scfenergies would reflect the dispersion correction. Consider, if I do a DFT+D (for whatever dispersion correction), the intuition would be that scfenergies would reflect the end of that calculation. Otherwise, I’d have run a DFT calculation without dispersion.

I have a whole pile of energies that I need to go fix because we used scfenergies from ORCA calculations…