[Pw_forum] about PWscf PP generation

Emine Kucukbenli kucukben at sissa.it
Sun Sep 11 14:55:32 CEST 2011


Dear Robin,
It takes some trial and error to generate a good pseudopotential. the  
more complex your atom is, harder it may get. So why dont you try  
starting with something simple? It is really not easy to guess the  
best rcut values without carefully examining the atom.

If I may, i would suggest you to start with something simple, draw the  
atomic orbitals, study the shape of pseduo orbitals for each cut off  
you choose and change them systematically. Watch also the logarithmic  
derivatives for ghosts. if you are generating a paw one, also watch  
the behavior of partial wave expansion curves. Do this for a simple  
system, actually, do it for a system that you already know the working  
set of parameters, like the ones in pslibrary project that you can  
download from qe-forge..once you find a reasonable range for all the  
parameters you can start refining them for a better pseudo.
And once you have a better idea about how each parameter affects the  
pseudopotential in the simpler system, i think it gets a tad easier to  
move on to harder-to-generate pseudopotentials.
best,
emine kucukbenli, phd student, sissa, italy


Quoting Robin H <robinshy at gmail.com>:

>  I can't find the answer to my question in the manual,so I put my question
> on the forum.I never wanted to make troubles.I think maybe someone has ever
> been confused by this kind of problem,and if they have solved it,could
> please to share the method with me.I 'm so appreciate it.It's much better to
> discuss than to work alone in the office.
>



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





More information about the users mailing list