[Pw_forum] Symmetry-finding problems in pw.x

[Nicola Marzari]

Hi Steve,


here are a few comments (in your case, 2), and then 1) are the likely
culprits).

1) you need to have enough significant digits in your
input for the code to recognize symmetry operations - 5 is probably
borderline, and not enough.

2) you shouldn't force to code to look for fractional translations:
arrange the nanotube such that your coordinates show
symmetry: x in -x, or such (that is to say, place the nanotube so that 
it is bisected by symmetry planes such as x=0).

You should also start with a SWNT, and experiment a bit to see when
operations are found.

3) for an ultrasoft carbon, you need between 25 and 30 Ry. your cutoff 
is much much larger than needed or reasonable. But you need a dual of
8 to 12 for the charge density cutoff. The tests are actually published 
in the pseudopotential page, for this pseudo.

Advice: try to reproduce diamond and graphene results, and then
a simple swnt, before moving to mwnt. Extensive results on
carbon, with this pseudo, are e.g. on a Mounet/Marzari PRB of 2005.

A question for everyone else: could there be an input mode in which
point-group symmetry operations are specified by hand ? Once could have
the code then 1) check that those are symmetry operations, or 2)
multiply atoms using the point group symmetries that have been read.


			nicola



Steven Kirk wrote:

> Hello,
> 
> I am trying to simulate a multiwalled carbon nanotube in pw.x using the
> input file:
> 
>   &CONTROL
>    title='MWNT calculation',
>    calculation='scf',
>    restart_mode='from_scratch',
>    prefix='MWNT',
>  /
>  &SYSTEM
>    ibrav = 6,
>    A = 17.0, B = 17.0, C = 4.26, cosAB = 0, cosAC = 0, cosBC = 0,
>    nat = 24,
>    ntyp = 1,
>    ecutwfc = 73.4986,
>  /
>  &ELECTRONS
>  /
> ATOMIC_SPECIES
> C 12.0107 C.pbe-rrkjus.UPF
> ATOMIC_POSITIONS {crystal}
> C     0.18067    0.03594    0.47051
> C     0.18067    0.03594    0.80385
> C     0.15316    0.10234    0.97051
> C     0.15316    0.10234    0.30385
> C     0.10234    0.15316    0.47051
> C     0.10234    0.15316    0.80385
> C     0.03594    0.18067    0.97051
> C     0.03594    0.18067    0.30385
> C    -0.35480   -0.09923    0.18505
> C    -0.32863   -0.16654    0.35171
> C    -0.35480   -0.09923    0.85171
> C    -0.28982   -0.22745    0.18505
> C    -0.23988   -0.27962    0.35171
> C    -0.32863   -0.16654    0.68505
> C    -0.28982   -0.22745    0.85171
> C    -0.18072   -0.32105    0.18505
> C    -0.11461   -0.35014    0.35171
> C    -0.23988   -0.27962    0.68505
> C    -0.18072   -0.32105    0.85171
> C    -0.04410   -0.36577    0.18505
> C     0.02810   -0.36734    0.35171
> C    -0.11461   -0.35014    0.68505
> C    -0.04410   -0.36577    0.85171
> C     0.02810   -0.36734    0.68505
> K_POINTS gamma
> 
> I am trying to make maximum use of symmetry, because the wavefunction
> file gets too big when I don't use it.
> 
> Unfortunately it seems that pw.x says 'No symmetry !' with the input
> data above. How can I make sure that pw.x detects the 4-fold symmetry
> and simulates the structure correctly with the minimum number of atoms ?
> 
> Many thanks in advance,
> Steve Kirk

-- 
---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu