CC question

I have a question for the experts in CC.

Which is the better set of natural orbitals those obtained from CCSD/cc-pVTZ or from CCSDT/cc-pVDZ ? 
This is in the context of non-covalent interactions. 
 

Hello!

Very hard to give advice without knowing any details. If I try to guess what you might need, I would suggest CCSDT/cc-pVDZ.
Best wishes,
Peter

Your guess is what I had in mind.

I want to perform a charge decomposition analysis between two fragments. Energies are not so important. What I need is a good quality wavefunction so that electron transactions are properly accounted for.

Triple zeta is beyond the computational effort that I can afford hence the question.

The answer depends on the system. If your system is the He atom or the Li+ ion, then CCSD and CCSDT are exactly the same (they are both equivalent to FCI for 2 electrons). So the bigger basis set will provide you more accuracy than the higher-order coupled cluster method.

In general though, DZ basis sets are not considered very accurate on their own.

I don't know what your system is, but in my applications it would be more accurate to do CCSD/cc-pVDZ and CCSD/cc-pVTZ and then do a (DZ,TZ)->CBS extrapolation, than to do CCSDTQP/cc-pVDZ or even FCI/cc-pVDZ on its own.

However, since you say that TZ is beyond the effort that you can afford, everything that I said is moot. You can do CCSDT/cc-pVDZ and say that you didn't try CCSD/cc-pVTZ because it was too expensive.

Also, since this question isn't about MRCC, you might find more answers at MMSE: mattermodeling.stackexchange.com/ with the "coupled cluster" tag. There's now more than 6100 users there!

What I have is something more in the real world such as CH3Br...NH3.

In addition as I've said I'm not looking for energies, I want to examine the chemical bond.