The ABINIT Users Mailing List ( CLOSED )

["Jess Kondor" <kondor.jess@gmail.com>]

 Dear Xavier and Andrew,

   Thank you very much for the suggestions.

    I realized that I used ecutsm (0.5 Ha) only for cell relaxation, the properties I got were obtained from static calculation (no ecutsm). As a result I have a high pressure. If I use ecutsm in static calculation I have a low pressure. So, should I still use ecutsm in all static calculations (after cell relaxation) ?

  Also, is the value 0.5 Ha for ecutsm OK? Somewhere I heard that in the Bernasconi method the value of ecutsm should be ~ ecut_eff/10 (~2.7 Ha for ecut=55 Ry).

 Thanks,
    Jess

On Dec 29, 2007 10:35 AM, Andrew Rappe <rappe@sas.upenn.edu> wrote:

Dear Jess and Xavier,

I agree with what has been said.
I would make a couple of points in addition:

1. If you have a high pressure, this could be due to a small change in
  lattice parameter.  Once you choose a plane-wave cutoff, you should
  relax your bulk system to find the theoretical equilibrium lattice
  parameter for that plane-wave cutoff.

2. You should use this same lattice parameter (found as above) as your
  in-plane lattice parameter.  If you don't follow these steps, the
  surface relaxations are likely to be quite different from those of
  the experimental system.

3. k-points are another important convergence parameter.  You should
  use the same in-plane k-point mesh in your bulk and surface cells.
  The only difference should be the use of a single sheet of k-points
  for the surface calculation, and a dense set of sheets of k-points
  for the bulk cell.  This way, the surface cell will have a similar
  in-plane stress state to the bulk cell at the same lattice
  parameter.

4. Finally, ecutsm is probably fine but also probably not needed for
  OPIUM psps because the energy convergence is often quite good.

Andrew

On Wed, Dec 26, 2007 at 11:18:55AM -0600 or thereabouts, Jess Kondor wrote:
> Dear Xavier,
>
>  Thanks for the prompt reply.
>
> >
> > Concerning the latter, do you use ecutsm ?
> >
>
> Yes, I do use ecutsm (0.5 Ha)
>
> >
> > > So, I am wondering which ecut I should use in my further
> > > calculations. I am going to study surfaces of my material.
> > >
> > You can likely explore your system with 45 Ry, then ... come back to
> > a convergence study
> > for the properties of the surface that you care about. Convergence
> > tests should
> > be done on the property you are interested in.
> >
> > X.
> >
> >
> Thanks for the suggestions,
>    JK