[Pw_forum] Single point energy calculation within PWSCF.

Emine Kucukbenli kucukben at sissa.it
Wed May 25 20:53:54 CEST 2011


Dear Hongsheng Zhao

> 2- If I can run a series of single-point energy calculations w.r.t
> different cut-off energies, k-point samplings, fine grids, cell size,
> and etc. Then look for when the energies, forces and stresses are all
> converged adequately.  In this way, I'll achieve the same aim as
> 'vc-relax' done for me, am I right?  I mean, if the energy, forces and
> stresses are all accurately converged based on a series of single-point
> energy calculations, then the final structure will also be accurate.

I was thinking that you were trying to optimize the calculation  
parameters however with this post I am lost.I will give a generic  
answer:

Before starting to work on a system, you need to optimize the  
calculation parameters, much like an experimentalist optimizing their  
machine set up before an experiment.

You need to make sure that what you will get, once you do the desired  
calculation, (like a desired experiment) would be sensible data. This  
is a preparation scheme _before_ the actual calculations.

In this preparation process, you need to check the convergence of  
energy, and/or forces, pressure etc, basically whatever interests you,  
or whatever the interested property is sensitive to. And by  
convergence, I mean convergence with respect to calculation  
parameters: ecutwfc/rho ; kpoints ; smearing for metals; box size for  
isolated systems, convergence thresholds etc ..again, whatever is  
relevant.

Once you know the optimum values for these parameters, you can do some  
applications: One of them might be finding the equilibrium lattice  
parameter. If  so, now you can run a "vc-relax" calculation to get the  
zero pressure structure..

Hope this made any sense and didnt confuse you even more.
cheers,
emine kucukbenli, phd student, sissa, italy


----------------------------------------------------------------
   SISSA Webmail https://webmail.sissa.it/
   Powered by Horde http://www.horde.org/




More information about the Pw_forum mailing list