[Pw_forum] Occupations in CP

Giovanni La Penna glapenna at iccom.cnr.it
Wed Sep 15 15:19:53 CEST 2010


Dear all,

I tried to perform CP-MD simulations moving
electrons beyond the highest KS state.
In theory, this can be done using

occupations='from_input'

and playing with the OCCUPATIONS card, using
a number of bands larger than half of the number
of electrons.
Here is an example for isolated H2 molecule:


# electron steepest-descent
  &control
     calculation='cp',
     restart_mode='from_scratch',
     prefix='elsd2',
     pseudo_dir='/home/lapenna/espresso_pseudo/',
     outdir='/home/lapenna/deloc/H2/./tmp/',
     ekin_conv_thr =1.e-06,
     nstep=50, iprint=10, isave=100,
     dt=2.0,
     ndr=50, ndw=60,
  /
  &system
     ibrav= 1,
     celldm(1)= 11,
     nat=2, ntyp=1,
     tot_charge=0.,
!    nelec=2,
     nspin=1, nbnd=2,
     ecutwfc = 25.0, ecutrho = 250.0,
     nr1b = 20, nr2b = 20, nr3b = 20,
!    occupations='fixed',
     occupations='from_input',
  /
  &electrons
     emass = 400.,
     electron_dynamics = 'sd',
!    electron_dynamics = 'damp',
     orthogonalization = 'ortho',
     ortho_max = 1000,
  /
  &ions
     ion_dynamics = 'none',
     ion_positions = 'from_input',
!    ion_dynamics = 'damp',
  /

ATOMIC_SPECIES
  H 1.008  H.pbe-rrkjus.UPF

ATOMIC_POSITIONS {bohr}
H    0.472420E+01  0.472438E+01  0.476847E+01
H    0.472443E+01  0.472425E+01  0.619194E+01

OCCUPATIONS
0.0 2.0


In the output (QE-4.2) I read:

...
    Wave Initialization: random initial wave-functions
    Occupation number from init
    nbnd =     2
     0.00 0.00
...

Same occupation numbers every time they are printed,
inconsistently with the charge of the system that
is zero.
Continuing with the dynamics, the H2 dissociation occurs
as expected, while with occupations 2. 0. the correct
minimum is satisifed. Maybe the comparison between
different occupations in CP-MD is nonsense, but this is another
story.

Why the occupations are all zero beyond the last non-zero
from input?
Is it a safety measure preventing such hazardous experiments?
Also in this latter case, the measure does not work,
except in the case of a careful output reading...

PW prints occupations as in input.

Thank you in advance,

                Giovanni

============================================================
Giovanni La Penna - National research council (Cnr)
Institute for chemistry of organo-metallic compounds (Iccom)
via Madonna del Piano 10,
I-50019 Sesto Fiorentino, Firenze, Italy
tel.: +39 055 522-5264, fax: +39 055 522-5203
e-mail: glapenna at iccom.cnr.it - http://www.iccom.cnr.it/lapenna
skype: giovannilapenna
============================================================


More information about the Pw_forum mailing list