Dear Lukas,
Thank you for your interest in the method!
You can find example inputs in the MTEST folder, for instance:
MINP_H2O_aug-cc-pVDZ_RS-PBE-P86_SOS-ADC(2)_direct
The main issue with your input is that you should use the combination calc=TDA and dft=RS-PBE-P86 instead of calc=RS-PBE-P86.
Please also consider the following points:
1. SCF initial guess: If you use scfiguess=restart, please make sure that the appropriate guess is available.
2. SCF acceleration: RS-PBE-P86/SCS-ADC(2) calculations are relatively expensive, so the computational cost of the SCF solution is likely negligible compared to the overall calculation. Therefore, I would recommend disabling the keywords intended to accelerate the SCF procedure, such as occri, fmm, and pssp. Their compatibility with excited-state calculations has not been tested. Alternatively, you could perform the ground-state calculation separately and then restart the excited-state calculation with scfiguess=off, ensuring you retain the necessary files from the ground-state calculation.
3. ADC(2) settings and spin scaling: I would recommend removing the dhexc=SCS-ADC(2) keyword, since the ADC(2) ansatz is used by default. However, the spin-scaling factors must be specified explicitly. Unfortunately, this is not entirely clear. According to the paper DOI: 10.1021/acs.jctc.1c01100, the appropriate factors are scsps=1.24 and scspt=0.64.
calc=TDA
dft=RS-PBE-P86
scsps=1.24
scspt=0.64
Please feel free to contact me if you have any further questions or encounter additional issues.
Best regards,
Dávid
