[CPMD-list] Output of TDDFT geometry optimisation

[Juerg Hutter]

Hi

>
> I see. This is related to the use of the "PROJECT" keyword in the
> "&CPMD ..  &END" section, isn't it ?

No, this keyword only affects the wavefunction gradient, not the
nuclear gradient in geometry optimization.

>
> Ok. Thus, the calculation of the PBE ground-state energy
> does not rely on KS energies and wavefunctions that would
> have been obtained with a xc potential other than the PBE
> potential, (for any given reason). That is, the ground-state
> and the TDDFT calculations are totally separated.

Yes, that's how linear response calculations work.

>
>
>
> Finally, there actually exist two ways to evaluate excited-state
> nuclear gradients in CPMD: yours, JChemPhys_118_3928; and
> that of Doltsinis and  Kosov, JChemPhys_122_144101. Which
> one is being used ?
>

Only the version from the first paper is implemented in CPMD.

regards

Juerg Hutter