help, DFT + U affect the parameters of the cell !

[Matthias Krack]

Dear Mohamed,

as you guessed, if you want to keep the cell vectors fix, then you have to 
choose the RUN_TYPE GEO_OPT instead of CELL_OPT. In this case only the 
atomic positions are relaxed.

There is the option to smear the MO occupations in general using
http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/SMEAR.html
which does not specifically affect the smearing of the Zn 3d orbitals.
Alternatively, you may use
http://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/SUBSYS/KIND/DFT_PLUS_U/ENFORCE_OCCUPATION.html#desc_SMEAR
which will specifically perform a smearing within the orbitals of the 
actual atomic kind affected by the DFT+U method.

Matthias



Matthias

On Wednesday, 25 June 2014 14:09:22 UTC+2, mohamed khuili wrote:
>
>
> Thank you Mr. Krack
> I began by optimizing the cell, using "CELL_OPT" and supercelle of 3X3X3, 
> I got good results vector (a) and (b) and (c) "alpha" "beta" and " gamma 
> ", but I get a gap 0.57ev smaller than that of the experimental 3.4 ev. and 
> the littérarure this is normal that the DFT underestimates the binding 
> energy of the Zn-d semi-core states and Applied sowe-have the DFT + U, I 
> redid the same calculation, I used "RUN_TYPE CELL_OPT "I got close to the 
> experimental gap, but I lost my optimized abc alpha beta and gamma 
> parameter:
> a = 3.259 became 3.0
> b = 3259 became 3.0
> c = 5.22 became 4.9
> the question is:
> - How can I apply DFT + U without losing vleurs my vectors I got on DFT?
> - Is what I should use "RUN_TYP GEO_OPT" instead of "RUN_TYPE CELL_OPT"
> I met Mr. David Benoit University of HULL, in Rabat in Morocco in the 
> project "CAPZEO" and if I understood what he told me, he suggested I try 
> with the "* smearing of the MO occupation*" section to 
> relocate(delocalised) the electron orbitals of the 3d of Zinc, but I did 
> not come to do
> I ask you, Mr Matthias to help me solve this problem
> Thank you
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140625/7c101669/attachment.htm>