CCSD (etc) 1-RDM properties not reproduced via MOLDEN + CCDENSITIES

Hi all,

I've been running CCSD (and CCSDT) calculations with the goal to reuse the 1-RDM. As far as I can tell, the file CCDENSITIES contains the elements of the 2-RDM (I ignore) and 1-RDM in the original ROHF basis.

When I read these files into pyscf I can reproduce 1-RDM properties (<T> and <V>) more or less exactly, via the MOLDEN read interface in pyscf. But the CCSD variants are out by up to 40 mHa. Both in comparison to direct outputs of MRCC.

I don't think I'm misunderstanding the contents of CCDENSITIES, given how close my values are to the real deal. But I would very much appreciate help to learn what it is I am (clearly) missing.

Many thanks in advance,

Tim

PS. Also, thanks to MRCC for its ability to write 1- and 2-RDMs! It's super handy.

 

Dear Tim,
In the file CCDENSITIES, the unrelaxed MO CC density matrices are stored excluding the core contributions.
In the output of executable "prop", the properties computed from the relaxed densities including the core contributions are printed out. You can only reproduce them if you correlate all electrons,  run a full CC (= full CI) calculation, and then, contract the 1-RDM from CCDENSITIES with the appropriate property integrals.
The dipole moments calculated from the unrelaxed density matrices are printed out at the end of the output of executable "mrcc". You can reproduce them if you correlate all electrons and contract the 1-RDM from CCDENSITIES with dipole moment integrals.

Brilliant. I can confirm I reproduce the dipoles

Many thanks for your help!

Tim