CCSDT / CCSDTQ on BS-UHF Singlet Reference

Hello!

I was wondering if it was possible to do single reference CCSDT or CCSDTQ calculations with the MRCC on a singlet broken-symmetry UHF reference. My initial tests seem to run into an error, yielding CC energies of NaN. When using the below input removing the occupation keyword and switching scftype=rhf, everything seems to be working fine, and the CCSDTQ O2 files from the MTEST directory using scftype=uhf seem to complete and yield the correct energies, so it seems somehow to be related to the broken symmetry singlet, not just the UHF reference.

A sample input and output are given below:

Relevant part of output attached.

Thanks,
Greg

Dear Greg,
There is no "occupation" keyword, the relevant keyword is occ. Probably you used this and set occ=7,2/8,1. In this case, you will get a high-spin triplet state. For a broken-symmetry UHF solution, you should lower the symmetry using keyword cmpgrp or turn it off using symm=off. Even in this case, it is not guaranteed that you get the BS symmetry solution. Then you should try to run quadratic SCF (qscf=AugHessG).

Dr. Kallay,

Thank you for your kind response. Yes, I did use occ= as opposed to occupation (my error in transcription).

I suppose I spoke imprecisely. I'm looking for the UHF solution corresponding to an Ms = 0 with the symmetry given (which I understand is properly a mixture of singlet, triplet, and higher spin multiplicity states), and to do CCSDTQ on this reference. Do I still need to go through quadratic SCF in a symm=off calculation to do this?

EDIT: The SCF did converge with the energy I expected from other related calculations with CFOUR, so I think the SCF has converged to my reference of interest.