SOCME Implementation in TURBOMOLE

[imamchenko]

Dear all,

I have been testing the SOCME functionality introduced in TURBOMOLE 7.9 (using proper) and for the S(0) to T(n) couplings we consistently observe around a 20% difference between TURBOMOLE and PySOC results. We have tried different functionals and basis sets, and also checked whether the discrepancy could be due to TURBOMOLE using spherical Gaussians vs PySOC using Cartesian - but it persists even with a Pople basis set. This difference is not observed when comparing ORCA and PySOC results, which agree well with each other. The S(n) to T(n) couplings agree well across all three programs, so the discrepancy appears to be specific to the ground state.

I have not been able to find a clear description of the working equations implemented in TURBOMOLE for this case. If anyone knows what math TURBOMOLE actually uses for the ground-state SOCMEs in TD-DFT, or can point to relevant documentation, I would greatly appreciate it.

Thank you in advance,
Iryna

[chris.hol]

Dear Iryna,

Turbomole uses the full (X-Y) part from TD-DFT to calculate the S0-Tn SOCME contributions, which arise from a skew-symmetric operator. See equation 37 in https://pubs.aip.org/aip/jcp/article/161/24/244105/3328183 for the working equation in Turbomole, where we also specifically discuss this difference between the Turbomole ansatz and the ones used in other programs. For excited state Sn-Tm and Tn-Tm SOCMEs, Turbomole can also calculate the relaxed contribution using "$scfinstab socme full" as outlined in the paper, which also may yield larger deviations to other programs, but it's quite expensive to do so.

Best,
Christof

[imamchenko]

Dear Christof,

Thank you so much for the fast reply and for pointing to the relevant equation, this is very helpful!

Best,
Iryna