The ABINIT Users Mailing List ( CLOSED )

[Scott Beckman <spbeckman@gmail.com>]

Hi,

It is also worth noting that if you are performing structural  
relaxations you will want to examine the convergence of the forces.

If you plot the components of the force on an atom versus ecut you  
will observe (small) fluctuations in the force.

If you want to report that your structures were optimized until the  
forces were smaller than XYZ meV/Angstrom you need to pick an ecut  
that is sufficient for the forces to be trusted to this precision.

Scott

On Aug 3, 2007, at 2:53 AM, Anglade Pierre-Matthieu wrote:

> Hi
>
> There is no such "true" ecut:
> the cutoff energy describes how many plane waves wil be used to
> describe the wave functions within your calculations. The more plane
> waves the more frequencies will be depicted and then the more accurate
> will be your calculation.
> You have to find an equilibrium between the computation time and the
> accuracy of your calculation. That's why we usually search the minimum
> ecut that accurately describe the system. Here "accurately" may
> strongly depends on the properties you want to examine. If you have no
> special requirement an absolute convergence of about 1e-3 Ha is
> usually considered correct because pseudopotentials tends to reproduce
> the full electron calculation on which they are based to this
> precision. That is, you will (almost) never get more physics by having
> something better converged. However some problems may require
> convergency of ecut down to 1e-4 or 1e-5 Ha.
>
> You can find examples of how to find the correct value of ecut in the
> Abinit tutorials. HAve you tryed them and reproduce the steps depicted
> there in your calculations ? If you want some comments on your own
> results, please send some details about your input and output.
>
> Also,  I'm just back from a summer school where I have presented
> Abinit to very beginners in DFT - PW calculations. Then, I have lots
> of didactical materials where I insist on this very problem. If you
> like I can send those PDF to you.
>
> regards
>
> pierre-matthieu Anglade
>
> On 8/2/07, m_mousavi1980@yahoo.com <m_mousavi1980@yahoo.com> wrote:
> > Hi
> >
> > I have a question
> > How can i understand which ecut for zno is true.
> > I get ecut 62.01
> > thanks
> > best regard
>
>
> --
> Pierre-Matthieu Anglade