[Pw_forum] modified Hartree exchange potential

[Paolo Giannozzi]

On Jul 5, 2007, at 15:30 , Lorenzo Paulatto wrote:

>> I want to modifiy the Hartree exchange potential to take into  
>> account a
>> screened coulomb potential. Which files need to be altered? I  
>> noticed that
>> a subroutine hartree(..) is called but couldn't find it.
>
> it is v_h(..) in PW/v_of_rho.f90; remember that the potential you  
> use in
> the scf code must be consistent with the one you have used in the  
> atomic
> (ld1) code to generate the pseudopotential. The latter is computed in
> atomic/hartree.f90

also note that the scf code implicitly assumes the 1/r dependence
when calculating the G=0 term of the potential

>> Also I wanted to change the exchange functional for a new functional.
>> I changed Modules/functionals.f90 to call a new functional, iexch=8,
>> and then added the new functional in flib/functionals.f90 as a new
>> subroutine under the slater one.
>> [snip]
>> Is this sufficient, or are there any other routines I need to change?

Routines "set_dft_from_name" and "_from_indices" should be modified
accordingly

> Same as above: the atomic code must be consistent, otherwise  
> unpredictable
> things *may* happen.

In principle, pseudopotentials should be generated using the same  
functional
that is used in the scf calculation, but nothing dramatic will happen  
otherwise.

In any event: it is wise to try new functionals on atoms first. The  
atomic code
uses the same routines as the scf code for functionals.

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy