[Pw_forum] Pw.x : relaxation in crystall coordinates?

Alexander Shaposhnikov shaposh at isp.nsc.ru
Fri Dec 16 12:17:39 CET 2005


Here is (minimum) tests job:
////////

 &CONTROL
 calculation = 'relax' ,
 restart_mode = 'from_scratch' ,
 outdir = '/scr/tmp/zro2t' ,
 pseudo_dir= '/home/alex/programs/espresso-current/pseudo/',
 prefix = zro2t  ,
 tprnfor = .true.,

 /
 &SYSTEM
 ibrav = 6,
 celldm(1) = 6.88,
 celldm(3) = 1.448,
 nat = 6,
 ntyp = 2,
 ecutwfc = 55,
 /
 &ELECTRONS
 diagonalization = 'david',
 /
 &IONS
 ion_dynamics='bfgs',
 /
ATOMIC_SPECIES
  Zr   91.00000  Zr.pbe-nsp-van.UPF
  O    16.00000  O.pbe-rrkjus.UPF
ATOMIC_POSITIONS {crystall}
    Zr     0.750000000    0.250000000    0.250000000    0  0  0
    Zr     0.250000000    0.750000000    0.750000000    0  0  0
    O      0.250000000    0.250000000    0.454600000    0  0  0
    O      0.750000000    0.750000000    0.545400000    0  0  0
    O      0.250000000    0.250000000    0.954600000    0  0  0
    O      0.750000000    0.750000000    0.045400000    0  0  0
K_POINTS {automatic}
4 4 3 0 0 0
//////////
Atomic positions in alat :
        Zr   0.7500000   0.2500000   0.3620000
        Zr   0.2500000   0.7500000   1.0860000
        O    0.2500000   0.2500000   0.6582608
        O    0.7500000   0.7500000   0.7897392
        O    0.2500000   0.2500000   1.3822608
        O    0.7500000   0.7500000   0.0657392

In the case of {crystall} coordinates, total force is always zero.

Regards,
Alex


On Thu, 2005-12-15 at 16:38 +0100, Paolo Giannozzi wrote:
> On Thursday 15 December 2005 14:50, Alexander Shaposhnikov wrote:
> 
> > is it possible to make structural relaxation with input coordinates
> > in "crystall" units? Total force always calculated = 0 and no
> > relaxation is done. The same system input in "alat" units (coordinates
> > taken from pw.x output) have non-zero total force.
> 
> the code prints cartesian coordinates, always. Please check whether
> you get different results with the same set of coordinates. If so, 
> please submit an test job
> 
> Paolo
> 




More information about the users mailing list