[Pw_forum] Calculations of magnetic systems by pwscf

[Stefano de Gironcoli]

In the new intup format for version 1.2.0 the options to be
specified in the &system namelist are:
  nspin=2  (that replace lsda=.true.)
  starting_magnetization(1)=0.9  this has to be specified for each
atomic type (if not specified it is assumed equal to zero) and defines
the magnetization used for  the atomic charges when the first starting
potential is built.

The value must be between -1 (totaly down magnetized) and +1 (totally
up magnetized).

If you want to do an AF material (NiO for instance) you must use different
data type for the  Ni up and down sites (obviously using the same
pseudopotential) with different values for starting_magnetization.
This is only the initial guess for the potential. There is no garantee that this
configuration is mantained but it usually does.

Stefano de Gironcoli


Eyvaz Isaev wrote:

> Dear Raghani,
>
> Unfortunately, I was not in the workshop and I don't
> know what kind of reasons were provided for this.
> Nevertheless, I did a lot of calculations for magnetic
> systems and results were good.
>
> So, you should add the next lines into your input
> file:
>
> lsda=.true.,
> starting_magnetization(1)=0.9,
>
> where the index means the magnetization for a given
> sort of atoms.
>
> Regards,
> Eyvaz.
> Theoretical Physics Department
> Moscow State Institute of Steel and Alloys
>
> --- Raghani Pushpa <pushpa at jncasr.ac.in> wrote:
> >
> > Dear all,
> > Can we do the calculations on magnetic systems by
> > using pwscf? I heard in
> > the workshop that pwscf can not handle the magnetic
> > systems but in the
> > code there is the provision for lsda calculations!
> >
> > Thanks,
> > Pushpa
> > --