ionic relaxation and PES in the excited state (TDDFT or GW/BSE)?

[awang95]

Hi,

I'm wondering, is there any way to do ionic relaxations and NEB calculations in the excited state (i.e. via TD-DFT or GW/BSE), in other words get a potential energy surface for excited states? If so, what would be the appropriate INCAR tags to do so? I've tried my best to figure out but am having trouble.

thanks!

[henrique_miranda]

To the best of my knowledge, the best way to do such calculations would be by freezing the occupations of an excited state using ISMEAR=-2 and FERDO.
The difficulty would be to ensure that the excited occupations you write in the INCAR remain valid for the different ionic configurations.
During relaxation, the order of the electronic levels might change which might lead to unexpected results.

Since you want to do an ionic relaxation and NEB you need a method for which the computation of forces is implemented.
In VASP you can do it for GGAs, Meta-GGAs, and Hybrid functionals.

TD-DFT is only implemented in VASP for the computation of the optical spectra through fxc. You cannot use it to compute forces which you would need for ionic relaxation and NEB.
Also for GW/BSE you don't have access to forces.

In principle, you can do it using RPA but would require careful testing.
My recommendation would be to start with the simpler approaches.

[awang95]

ok thanks for explaining!