[Q-e-developers] smearing style

[stefano de gironcoli]

dear Jiqiang Li,
   the VBM is probably degenerate and depending of minute details (like 
the smearing) which state comes last can be randomly assigned.
   It would make more sense to compute some density of state related 
quantity adding contribution from all states (weighted with
a smeared delta-function)   \sum_i \delta(e_i-e_VBM)|<f|dVtot/dR|i>|^2 / 
\sum_i \delta(e_i-e_VBM).
   i think this should be more stable.

   there are options in the pp.x tool to plot a single wavefunction or 
the locally density of states at the Fermi energy. This could be easily 
modified to compute ldos(e_VBM) as a function of e_VBM as a parameter. 
And you could see if the problem is related to degeneracy.

  regards,
stefano

On 20/09/2015 04:29, Jiqiang Li wrote:
> Dear Stefano,
>        The system is a semiconductor, but it is +1 charged. Pwscf 
> forces me to set a value for degauss. I want to evaluate the 
> <f|dVtot/dR|i>, where the |i> and |f> refer to the wavefunctions of 
> valence band maximum (VBM) and the isolated state induced by defect in 
> the band gap, respectively. I found that the related wavefucntions 
> change a lot when the weights (occupations) of the highest bands 
> change. By the way, I am performing a Gamma point only calculation for 
> a large supercell with a lattice constant exceeding 10 Angstrom.
>         I believe the many quantities (such as total energy) do not 
> change significantly when the occupations of highest states vary, 
> because contribution from these states is relatively small than that 
> from the rest states. So I am wondering whether it is normal that 
> wavefunctions of highest occupied states change remarkably when the 
> occupations of highest states vary.
>
> Best regards!
> Jiqiang
>
> ------------------------------------------------------------------------
>
>     *From:* stefano de gironcoli <mailto:degironc at sissa.it>
>     *Date:* 2015-09-20 04:07
>     *To:* q-e-developers <mailto:q-e-developers at qe-forge.org>
>     *Subject:* Re: [Q-e-developers] smearing style
>     Dear Jiqianq
>      the answer to your question depends on which physical quantity
>     you want to calculate.
>      Is the state you want to study spatially localized ? is it an
>     isolated (narrow band) in a band energy gap ? are you interested
>     in the shallow donor states formed by the holes at the bottom of
>     the conduction band of a doped semiconductor ? Or is your system a
>     metal ?
>      clarifying to yourself what is the question you want to answer
>     will help you to figure out how to compute a physically meaningful
>     number.
>
>     stefano
>
>     On 19/09/2015 12:37, Jiqiang Li wrote:
>>     Dear Lorenzo,
>>             For example, for a supercell with 100 electrons, it will
>>     be of 99 electrons when the system is +1 charged. I consider the
>>     50th band as the highest occupied band with half occupation.
>>
>>     Jiqiang
>>
>>     ------------------------------------------------------------------------
>>
>>         *From:* Lorenzo Paulatto <mailto:paulatz at gmail.com>
>>         *Date:* 2015-09-19 18:14
>>         *To:* General discussion list for Quantum ESPRESSO developers
>>         <mailto:q-e-developers at qe-forge.org>
>>         *Subject:* Re: [Q-e-developers] smearing style
>>         Dear Jiqiang,
>>         the "highest occupied banda" are obviously defined only for
>>         insulators, i.e.
>>         there is no such thing as a highest occupied bands when using
>>         a smearing, what
>>         threshold do you use for occupied/non occupied?
>>         kind regards
>>         On Saturday, September 19, 2015 05:48:09 PM Jiqiang Li wrote:
>>         > Dear developers,
>>         >         I want to use the wavefunctions of highest occupied
>>         bands  in a
>>         > charged system to evaluate some quantities, such as
>>         psi_i(r) ＊psi_j(r) .
>>         > However, I found that the wavefunction sensitively depends
>>         on the degauss
>>         > value.  So, how to choose a suitable value for  degauss
>>         when I want  to
>>         > obatain a physically meaningful quantity.
>>         >
>>         > Best regards!
>>         >  Jiqiang Li
>>         >
>>         >
>>         >
>>         -- 
>>         Lorenzo Paulatto - Paris
>>         _______________________________________________
>>         Q-e-developers mailing list
>>         Q-e-developers at qe-forge.org
>>         http://qe-forge.org/mailman/listinfo/q-e-developers
>>
>>
>>
>>     _______________________________________________
>>     Q-e-developers mailing list
>>     Q-e-developers at qe-forge.org
>>     http://qe-forge.org/mailman/listinfo/q-e-developers
>
>
>
> _______________________________________________
> Q-e-developers mailing list
> Q-e-developers at qe-forge.org
> http://qe-forge.org/mailman/listinfo/q-e-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20150930/99bd7a47/attachment.html>