CBS extrapolation of CCSD(T) energies with TCE

[Stephen Weitzner]

Dear NWChem developers,

I am working on some CCSD(T) calculations and I would like to extrapolate the energies to the CBS limit. I am separately extrapolating the reference HF energy and the correlation energy. When looking at the TCE output, there is a listing of correlation and correction energies:

 CCSD[T]  correction energy / hartree =        -0.033456704628325
 CCSD[T] correlation energy / hartree =        -1.181158531598549
 CCSD[T] total energy / hartree       =      -312.821576078952774
 CCSD(T)  correction energy / hartree =        -0.032495461486941
 CCSD(T) correlation energy / hartree =        -1.180197288457164
 CCSD(T) total energy / hartree       =      -312.820614835811341

When doing CBS extrapolation, should I add the correction energy to the correlation energy prior to the extrapolation (e.g.,  -0.032495461486941 + -1.180197288457164), or should I only extract the reported CCSD(T) correlation energy (-1.180197288457164) to use in extrapolation?

Can you please also help to clarify what the CCSD(T) correction energy reported here actually is? I see good agreement between the CCSD(T) correlation energy printed above here with the CCSD(T) correlation energy I have obtained using the CCSD module.

Thanks,

Stephen

[Stephen Weitzner]

Apologies, I think I just misunderstood the naming convention being used here. It seems that the "CCSD(T)  correction energy" is the contribution from the perturbative triples and the "CCSD(T) correlation energy" is the sum of the CCSD correlation energy and the correction energy.