Dear Marcos,

Sorry for making a mistake for they are two different psp's.

Yes,we indeed can add some code to ATOM,but I do not familiar with the ATOM 
code,and only can have a lilttle use.

Now I doubt whether upf and psf have the same contents in them.If so,it makes 
the modifying the code easy,if not,...

Now I want to compare transiesta and ATK.But they have different foramt psp's.

Many people attempted to do this thing but I see no answers.

If I succeed in it ,I will inform this mail-list.

I think APE can solve this problem.


Yours Guangping

2010-09-29



Guangping Zhang



发件人: Marcos_Veríssimo_Alves <[email protected]>
发送时间: 2010-09-29 21:35
主 题: Re: [SIESTA-L] about the upf and psf generating from ATOM
收件人: [email protected]



Guangping,


Your header for th UPF pseudo is C, while that of psf is N...


Nevertheless, the way to proceed, for the moment, would be to take a look at


http://www.pwscf.org/wiki/index.php/Unified_pseudopotential_format


where the UPF format is fully specified. Then you could write a converter for 
the two formats. 


Otherwise, the cutoff radii of the UPF pseudo, including the nonlinear core 
correction radius, for the pseudos you have used in UPF format must be 
specified somewhere. Maybe in a website where you can download them: either 
specified explicitly or a citation to a publication that has used the cited 
pseudo.


The problem I see is that UPF needs the nonlinear core correction radius, and 
this is written nowhere in the psf file. You would need to calculate it by the 
information on how the pseudo core charge is determined given in the ATOM 
manual. Either you would deduce this or you could "cheat" by taking the nlcc 
radius directly from the .inp file, if you have it.


Either way, perhaps the easiest way to proceed would be to change the routine 
wrapup.f in ATOM, adding lines to write a full UPF file, besides the psf and 
vps ones. It would be nice if this were added to ATOM, because the UPF format 
is now read by several standard plane-wave programs and one painful thing is 
that whenever we have to check siesta calculations against PW ones, we never 
have the exactly corresponding pseudo.


Marcos


On Wed, Sep 29, 2010 at 4:53 AM, Guangping Zhang <[email protected]> wrote:

Dear siesta users:
Now I have to convert the pseudopotential upf format used in ATK into psf 
format used in Siesta or vice versa.
And I saw they are generated by the same progaram "ATOM",so there must be some 
way to convert them mutually.
The head of upf file:
---------------------------------------------------------
<PP_INFO>
 ATM3      no_date   Troullier-Martins                       
</PP_INFO>
<PP_HEADER>
    0                  Version Number
    C                  Element
   NC                  Norm - Conserving pseudopotential
    T                  Nonlinear Core Correction
 SLA  PW   PBE  PBE    PBE  Exchange-Correlation functional
                4      Z valence
                0      Total energy
  0.0000000  0.0000000 Suggested cutoff for wfc and rho
    3                  Max angular momentum component
 1006                  Number of points in mesh
    4    4             Number of Wavefunctions, Number of Projectors
 Wavefunctions         nl  l   occ
                       2S  0  2
                       2P  1  2
                       3D  2  0
                       4F  3  0
</PP_HEADER>
-----------------------------------------------------
while the head of psf file:
------------------------------------------------
 N  pb nrl pcec
 ATM3      12-MAR-06 Troullier-Martins                       
 2s 2.00  r= 1.24/2p 3.00  r= 1.24/3d 0.00  r= 1.24/4f 0.00  r= 1.24/  
   4  0 1018  0.354107453809E-03  0.125000000000E-01   5.00000000000    
 Radial grid follows
-------------------------------------------------------
So they are both generate by ATOM,but in siesta ,the ATOM does not have the 
functional to generate upf format.

Who can help me out?

Thanks in advance!

Yours Guangping
2010-09-29



Guangping Zhang

Responder a