[Octopus-users] Time units..

[Xavier Andrade]

Hi Nkululeko,

On Wed, 17 Jun 2009, Nkululeko Nokwara wrote:
> My first question today regards the units of time in a time-dependent run.
> When performing a time-dependent run after setting energy and length units
> to eV and Angstrom respectively (Units = eV_Angstrom), what units is time?
> What are the units of the time-step?

The units of time in this case are hbar/eV = 658.21189 attoseconds.

> Secondly, when checking the time-dependent parameters before perturbing the
> ground state, is it necessary to run the unperturbed time-dependent
> calculation for the complete set of intended time steps to determine the
> largest time step that one can use? What I mean is that when one wants to
> perform 5000 time-step calculations, should one check the unperturbed system
> (for constancy in the total energy) for all the 5000 time steps or a smaller
> number (50 say) suffice it, then perform the perturbed calculation for the
> 5000 time steps afterwards?

If you select a time step that is too large, the propagation will be 
unstable and it will explode in the first steps, so less than 100 steps 
where the energy is conserved are enough in most cases to check that the 
propagation is stable.

Cheers,

Xavier