[Pw_forum] Question on epsilon.x

[Andrea Ferretti]

Hi Mohsen,

the smearing technique is meant to be a numerical tool to perform the
thermodynamic limit (# of kpts going to infinity, i.e. size of the system going 
from being finite to extended).
The actual value of the smearing parameters thus depends on how far the 
evaluation of the limit is, i.e. on how many kpts you are using in the 
calculation. Generally, the bigger the # of kpts, the smaller the smearing
can be. The smaller the smearing, the higher the accuracy of the final limit.
But smearings that are too small (relative to the # of kpts) lead to unphysical 
features, especially in the spectra (due to the finiteness of the systems 
studied)...

After all, the value of the smearing (and the number of kpts) depend
on the accuracy you want to reach. (if you want eg to describe the 1/sqrt(E) 
divergences of a nanotube DOS and JDOS you probably need a lot of kpts).

A rule of thumb to avoid spurious features is the following:
given a finite mesh of kpts, the bands of the systems are discretized.
The smearing parameter should be of the order of the difference in the 
eigenvalues (transition energies in your case) corresponding to
nearby kpts (i.e., the smearing should be of the order of
the discretization induced on the eigenvalues, or transition energies
in the case of epsilon.x)

practically, I would start with the default value of the smearing, and then 
check a posteriori.

hope it helps
Andrea

> Dear Andrea,
> Thanks for your answer, I know the meaning of these parameters,
> I want to know how can i determine these parameters for a system
> (Like a CNT (6,0))?
> Thanks for your answers.
> Mohsen
> 
> On Mon, May 17, 2010 at 7:37 PM, Andrea Ferretti <ferretti at mit.edu> wrote:
> 
>
>       Dear Mohsen,
>
>       the meaning of these parameters (smearing for inter and intra band
>       transitions) is somehow explained in the epsilon.x manual,
>       ~espresso/Doc/eps_man.pdf
>
>       Andrea
> 
> 
> > Dear all,
> > I have not any answer for this question, Actually I have a problem
> > about eps.x, I can't understand this two parameters (intersmear,
> > intrasmear).
> > How can i determine (intersmear, intrasmear).
> > Thanks for your answers
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
> >
> 
> --
> Andrea Ferretti
> Oxford University, Department of Materials
> Parks Road, Oxford OX1 3PH, UK
> Tel: +44 (0)1865 612796;  Skype: andrea_ferretti
> URL: http://quasiamore.mit.edu
> 
> Please, if possible, don't send me MS Word or PowerPoint attachments
> Why? See:  http://www.gnu.org/philosophy/no-word-attachments.html
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 
> 
> 
>

-- 
Andrea Ferretti
Oxford University, Department of Materials
Parks Road, Oxford OX1 3PH, UK
Tel: +44 (0)1865 612796;  Skype: andrea_ferretti
URL: http://quasiamore.mit.edu

Please, if possible, don't send me MS Word or PowerPoint attachments
Why? See:  http://www.gnu.org/philosophy/no-word-attachments.html