[Pw_forum] Change of occupation number and cut3d

Samuel Poncé samuel.pon at gmail.com
Sun Jul 31 15:49:43 CEST 2011


Dear Le,

I have normally 8 electrons in my system and define a charge of tot_charge =
+0.00046875. I then have 7.99953125 electrons in my system. I want to do
than so that I can make a finite difference calculation with respect to the
occupation number of my HOMO. As I'm only considering the gamma point of
diamond and as the HOMO is 3X degenerate at that point I decrease the
occupation number of the last 3 valences bands from 2.0 electrons per bands
to 1.99 electrons per bands.

The numbers that occupation numbers gives in the output file of PWSCF is the
occupation number (theoretically 2.0 for the valence band) weighted by the
symmetries of the k point grid.

*Actual behaviour:*
     number of k points=    8
                       cart. coord. in units 2pi/a_0
        k(    1) = (   0.0000000   0.0000000   0.0000000), wk =   0.0312500
        k(    2) = (  -0.2500000   0.2500000  -0.2500000), wk =   0.2500000
        k(    3) = (   0.5000000  -0.5000000   0.5000000), wk =   0.1250000
        k(    4) = (   0.0000000   0.5000000   0.0000000), wk =   0.1875000
        k(    5) = (   0.7500000  -0.2500000   0.7500000), wk =   0.7500000
        k(    6) = (   0.5000000   0.0000000   0.5000000), wk =   0.3750000
        k(    7) = (   0.0000000  -1.0000000   0.0000000), wk =   0.0937500
        k(    8) = (  -0.5000000  -1.0000000   0.0000000), wk =   0.1875000

And for the gamma point:

 k = 0.0000 0.0000 0.0000 (  1243 PWs)   bands (ev):

    -7.9264  13.7887  13.7887  13.7887  19.5000  19.5000  19.5000  27.9310

     occupation numbers
     0.0312   0.0312   0.0312   0.0312   0.0000   0.0000   0.0000   0.0000

*Expected behaviour*
     number of k points=    8
                       cart. coord. in units 2pi/a_0
        k(    1) = (   0.0000000   0.0000000   0.0000000), wk =   0.0312500
        k(    2) = (  -0.2500000   0.2500000  -0.2500000), wk =   0.2500000
        k(    3) = (   0.5000000  -0.5000000   0.5000000), wk =   0.1250000
        k(    4) = (   0.0000000   0.5000000   0.0000000), wk =   0.1875000
        k(    5) = (   0.7500000  -0.2500000   0.7500000), wk =   0.7500000
        k(    6) = (   0.5000000   0.0000000   0.5000000), wk =   0.3750000
        k(    7) = (   0.0000000  -1.0000000   0.0000000), wk =   0.0937500
        k(    8) = (  -0.5000000  -1.0000000   0.0000000), wk =   0.1875000

And for the gamma point:

 k = 0.0000 0.0000 0.0000 (  1243 PWs)   bands (ev):

    -7.9264  13.7887  13.7887  13.7887  19.5000  19.5000  19.5000  27.9310

     occupation numbers
     0.0312   *0.03109375*   *0.03109375    **0.03109375   *0.0000
0.0000   0.0000   0.0000

( The bold number corresponds to 1.99/64 )

Here is my input file:

&CONTROL
  calculation  = "scf"                     ! scf run
  restart_mode = "from_scratch"            ! "restart" form an interrupted
run
  prefix       = "Diamond"                 ! Basename for all binary files
  verbosity    = "high"                    ! Output more informations
  wf_collect   = .true.                    ! Collect wf in one single file

  pseudo_dir   = '$PSEUDO_DIR/'
  outdir       = '$TMP_DIR/'

  tstress      = .true.                    ! If true, print stress
  tprnfor      = .true.                    ! If true, print forces
/
&SYSTEM
  nbnd      = 8                            ! 4 valence and 4 cond.
  ibrav     = 2                            ! Correspond to fcc lattice
  celldm(1) = 6.66491891                   ! Lattice spacing in Bohr
  nat       = 2                            ! Number of atoms
  ntyp      = 1                            ! Number of different atomic type
  ecutwfc   = 100                          ! Wf cutoff in Ry
  ecutrho   = 400                          ! Density cutoff (4*ecutwfc for
nc psp and 9-10 for ultrasoft)
  tot_charge = +0.00046875                 ! 4-(63*4+3.97)/64
  occupations = "fixed"
/
&ELECTRONS
  diagonalization = "david"                ! Diagonalization method: "david"
or "cg"
  conv_thr        = 1e-16                  ! Energy accuracy threshold in Ry
  mixing_beta     = 0.5                    ! Charge mixing, reduce it if SCF
is unstable
/

ATOMIC_SPECIES
C  12.011  C.pz-vbc.UPF                        ! Label  |  mass  |
pseudopotential

ATOMIC_POSITIONS {bohr}
C  0.00  0.00  00.00
C  1.66622973 1.66622973 1.66622973

K_POINTS automatic
4 4 4 0 0 0

Best regards,

Samuel Poncé

2011/7/30 Duy Le <ttduyle at gmail.com>

> 2011/7/30 Samuel Poncé <samuel.pon at gmail.com>:
> > I did try to add the variable tot_charge = +0.00046875 because I have 64
> k
> > points and 4 valences band ( 4-(63*4+3.97)/64)
>
> Not sure what you were calculating. tot_charge is additional charge
> you add to your system.
> For example, if you have a system with 8 electron, and if you define
> tot_charge = +1, you will have 8-1=7 electron in the system.
>
> > However if I add the variable 'occupations = "fixed"' then I get the
> > following WARNING:
> >      WARNING: integrated charge=     8.00000000, expected=     7.99953125
> >
> > I off course get the exact same total energy as with tot_charge = 0 (this
> is
> > not the expected behaviour).
> >
> > I did try without the variable 'occupation' (by the way what is the
> default
> > value for occupation?) and also get the exact same total energy.
>
> This does not depend on occupation. Read manual about occupation for
> more details.
> If you do a simple math: 7.99953125+0.00046875 you will see a "magic"
> number.
>
>
> > I also do not understand why I don't have an occupation of 2.0 for all my
> > valence states?
> >
> >  k = 0.0000 0.0000 0.0000 (  1243 PWs)   bands (ev):
> >
> >     -7.9264  13.7887  13.7887  13.7887
> >
> >      occupation numbers
> >      0.0312   0.0312   0.0312   0.0312
> you should post your input for further advice.
> --------------------------------------------------
> Duy Le
> PhD Candidate
> Department of Physics
> University of Central Florida.
>
> "Men don't need hand to do things"
>
>
> >
> >
> > 2011/7/30 GAO Zhe <flux_ray12 at 163.com>
> >>
> >> Dear Ponce:
> >> I am sorry that I may not catch the right meaning of your first
> question.
> >> If you mean how to make extra or less charge for a model, you can use
> >> "total_charge" in &system namelist.
> >> If you wanna get information of wavefunction, you'd better read
> >> Doc/INPUT_PP. For the charge density at specified k-point and band, you
> can
> >> calculate |psi|^2 through pp.x. Moreover, if you want to obtain the
> charge
> >> density for all k-points on specified bands, you can summary them by
> pp.x,
> >> too. P.S.: the original code of pp.x can just allow 7 filepp, but you
> can
> >> change the variable, nfilemax, in PP/chdens.f90, then make pp.x.
> >> Good Luck.
> >> --
> >> GAO Zhe
> >> CMC Lab, MSE, SNU, Seoul, S.Korea
> >>
> >> At 2011-07-29 17:27:27,"Samuel Poncé" <samuel.pon at gmail.com> wrote:
> >>
> >> Dear Quantum espresso user/developers,
> >>
> >> I have two questions:
> >>
> >> 1) Is it possible in pwscf to change "by hand" the occupation number (
> and
> >> impose a charge compensation accordingly) and if so how to do it?
> >> I would like to do so because the sum over the different mode of the
> >> partial derivative of the phonon frequency with respect of the
> occupation
> >> number gives the zero point motion renormalization using Brooks theorem.
> I
> >> would like to do this by finite difference (hence doing by hand some
> >> calculation at different occupation number).
> >>
> >>  2) Is it possible to extract informations in text format form the
> >> wavefunctions? What is the right post processing tool to do so? Like
> having
> >> the value of the wavefunction for a defined k points and band index?
> >>
> >> Thanks you!
> >>
> >> Samuel Poncé.
> >> (phD student, Université Catholique de Louvain, Belgium).
> >>
> >>
> >>
> >> _______________________________________________
> >> Pw_forum mailing list
> >> Pw_forum at pwscf.org
> >> http://www.democritos.it/mailman/listinfo/pw_forum
> >>
> >
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
> >
> >
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20110731/b4f173f1/attachment.htm 


More information about the Pw_forum mailing list