Spin-flip CI input and EOM-SF irrep

Questions
#1

Hello everyone,

As of recently, we have begun working with spin-flip methods in our group to describe some diradicals (some have triplet and others singlet ground states, we would like to be sure our previous method identifies the right one) and the privilege has fallen upon me to discover how to invoke these methods in Q-Chem. I think I figured out the SF-XCIS and SF-TDDFT methods, but I am still having some problems with others and I hope you can help me with them.

The first one concerns the SF-CI methods (i.e. CIS, CISD, CISDT). I would like to invoke SF-CIS just as a test, but I only get triplet and quintet states with an input that contains:
$rem
EXCHANGE hf
UNRESTRICTED false
SCF_ALGORITHM DIIS_GDM
CIS_N_ROOTS 3
CIS_SINGLETS true
CIS_TRIPLETS true
SPIN_FLIP true
BASIS jul-cc-pVTZ
$end
Are there other keywords I should add? Additionally, how do I transform this into CISD or CISDT? I found in the manual it seems to be possible, but I couldn’t immediately find how and some tests didn’t go very well.

As a second question: how should I specify the number of states of the irreproducible representations for molecules with the D∞h point group, e.g. in EOM-SF-CC calculations? Maybe this is a trivial question, but I couldn’t quite figure it out for the ∞ point groups.

Thanks in advance for any info you can supply.

Kind regards, Jochen

#2

Dear Jochen:
As an introduction to EOM-CC method and examples including SF calculations, check out Sessions 09-11 from our recent workshop:

The supporting materials include detailed explanations of how to set the calculation and analysis of the output including symmetry.
The detailed overview of the SF methods can be found here:

For linear molecules, Q-Chem will use largest Abelian subgroup, so you need to figure out which irreps your transitions belong to. They would be A1 and the irrep to which the triplet reference belongs to.

SF-CIS and SF-DFT do not use symmetry, so there you just ask for 4 target states without specifying symmetry.

We do not support CI methods other than SF-CIS.

A cheaper alternative to EOM-SF-CCSD would be EOM-SF-MP2 (it uses MP2 amplitudes instead of CCSD).

Also, be careful with using restricted=true because the ROHF convergence can be iffy. Try first unrestricted to see what is the situation with spin-contamination and then proceed with caution if you really need ROHF.

Good luck!