The ABINIT Users Mailing List ( CLOSED )

["Matteo Giantomassi" <Matteo.Giantomassi@uclouvain.be>]

> Hi Matteo
>
> Thank you very much for your super fast reply. the keywords you told
> me does help me reduce the size of kss, and calculations much faster.
>
> But i still have a few more questions here:
>
> 1. How do i make sure the diagnolization is paralleled in the KSS
> step? I didn't see anything like: "P newkpt: treating     17 bands
> with npw=    4356 for ikpt=   1 by node    0"

 Dear Aaron,

 Just look at the log files produced by each node in the directory where
 you are running. Each node is reporting the particular k-point it's
taking care of.

>
> 2. How do I guarantee that i don't lose too much accuracy when I
> specify the value of nbandkss?

 If you use kssform=1 (direct diagonalization of the Hamiltonian)
 there's no accuracy loss: the Hamiltonian is constructed starting
 from the self-consistent density of the previous dataset. Hence, if the
density is
 well converged, you get the exact eigenstates of the Kohn-Sham Hamiltonian.

 If you use the conjugate gradient method you have to make sure that all
 bands included in the calculation (both empty and occupied) are sufficiently
 converged. In this case, indeed, you are using iterative procedures to
 find the lowest nbandkss eigenvectors of the self-consistent Hamiltonian
matrix.
 That's the reason why, in the second dataset, I'm using tolwfr to stop
the NSCF calculation.

>
> 3. continuous of last question, What factors decide the values of
> nband and nbandkss? I do know that the easiest and the slowest way is
> to increase the value little by little and see when it converges, but
> it's not physics, would you mind tell me how I should do that?

  You don't need to perform convergence studies during the generation of
the KSS file.
  Just generate a KSS file containing a sufficiently large and affordable
number of converged empty states.
  As a rule of thumb at _least_ 10*(number_of_occupied_states).
  These empty states are used to assemble the Green function and the
screened interaction.
  You have to check carefully the dependency of your results on the number
of states only
  during the calculation of the screening and the calculation of the GW
corrections

 Best Regards,
 Matteo Giantomassi