The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

On 12 Jul 2007, at 10:23, zmq.lyao@gmail.com wrote:

> Dear all,
>    When I do a RF calculation, I set nband=the number of occupied  
> bands (not including empty bands) for the convergence of d/dk  
> calculation. However,  if I also set the nband=the number of  
> occupted bands in the first step GS calculation,the calculated  
> dielectric constant(Anaddb analysis) is much smaller than that from  
> the GS calculation including empty bands. But the former is closer  
> to the experimental value. I don't know why the number of empty  
> bands have so big influence on the dielectric constant. Do I need  
> to include enough empty bands in the first step GS calcualtion?  
> Thanks in advance.
>
> regard,
> ZM

Could you send your input files ?
Thanks
Xavier