[Pw_forum] Request about the "real" total energy

[Stefano de Gironcoli]

the pw codes in the QE distribution calculate the total energy in the 
pseudopotential framework.
Pseudopotentials should be good at computing energy differences (at the 
forzen-core level) . 
If you wish a total energy to be compared with an all-electron 
(forzen-core)energy you need to calculate the formation energy  of your 
system relative to isolated atoms and then use the difference between 
the isolated atom computed by the pseudopotential and with an 
all-electron code to correct it. 
Within the QE dsitribution, you can perform both all-electron and 
pseudopotential calculation in the atomic case with the ld1.x code . 

stefano

steve ndengué wrote:
> Dear all,
>
> We have an enquiry about some results appearing in the output file of the pw.x run.
>
> We
>  are trying to perform computations on isolated atoms to try to get a 
> significance of the results given by the code. We have to recall that we
>  are new users of QE. And as far as we know it is not too much indicated
>  for such kind of computations but it is rather a first try to get some 
> understanding on the "behaviour" of the code.
>
> So the problem is 
> that when performing the computations on He for example we obtain the 
> correct ground state energy up to the Hartree-Fock limit. But when 
> applied to other atoms (Li, Na, Be, C, ...) the value obtained do not 
> correspond at all to the known value.
>
> We thus have 2
>  questions:
>
> 1- Why is it that the level of refinement obtained is
>  the Hartree-Fock limit; we may have expected a better accuracy? Is it 
> linked to the use of pseudo-potentials?
> 2- What can we do to recover the correct (up to the Hartree-Fock limit at least) total energy for the other atoms? 
>  
> Sincerely Yours,
> On behalf of the 
> Groupe  de Structure et Dynamique des Atomes et Molecules
> Laboratoire de Physique Fondamentale
> Département de Physique
> Université de Douala.
>
> Steve.
>
>
>       
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>