My Community

Hello,

I am getting big total drift along the Z-axis for a metal oxide cluster model. Why it gives big total drift alone the Z-axis?

Thanks,
Agalya

I am using the following input.

System =M1 phase optimization


ISTART=1

ICHARG=1

NPAR = 6

LPLANE = .TRUE.

LREAL = auto

!ROPT = 2e-4 2e-4

PREC = Low

ALGO = Fast
NSIM =1


NGX = 96
NGY = 96
NGZ = 70

!IDIPOL = 4

!NBANDS = 1630
!NELM = 100
EDIFF=1e-5

ISMEAR=0
SIGMA=0.01

ISPIN=2
NUPDOWN=1
!MAGMOM=2 2 2 2 2 2 2 2
!ISIF = 2


NSW=100
IBRION = 2
!POTIM = 0.4
EDIFFG=-0.05
!NFREE = 10

! write wavefunction
LWAVE=.TRUE.
! write chg and chgcar
LCHARG=.TRUE.


The total drift of the final optimized geometry is given below.
POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
13.74209 9.39535 7.20296 -0.025839 -0.015961 -0.026117
6.25791 10.60465 7.20296 0.025839 0.015961 -0.026117
12.54793 12.85129 7.19424 -0.017165 -0.006918 -0.007717
7.45207 7.14871 7.19424 0.017165 0.006918 -0.007717
9.00117 12.86621 7.60754 0.029310 0.000896 -0.013394
10.99883 7.13379 7.60754 -0.029310 -0.000896 -0.013394
10.00000 10.00000 8.20823 0.000000 0.000000 0.039782
5.86000 6.07300 7.55484 0.913071 1.715190 -0.065139
8.31050 10.91654 7.61178 -0.004646 0.007390 0.003120
6.38800 12.49800 7.62285 1.147898 0.011871 -0.225623
4.40400 10.67700 7.67184 1.281401 -1.153634 -0.103094
6.56409 8.79348 7.66373 0.017370 0.016439 -0.009962
13.72789 9.45242 5.49736 -0.001459 0.014796 0.009705
10.75358 11.77588 7.63714 0.022114 -0.001227 -0.012731
14.14000 13.92700 7.55484 -0.913071 -1.715190 -0.065139
13.43591 11.20652 7.66373 -0.017370 -0.016439 -0.009962
11.24700 14.24700 7.65885 -0.208736 -0.969166 -0.208129
8.75300 5.75300 7.65885 0.208736 0.969166 -0.208129
11.68950 9.08346 7.61178 0.004646 -0.007390 0.003120
13.61200 7.50200 7.62285 -1.147898 -0.011871 -0.225623
15.59600 9.32300 7.67184 -1.281401 1.153634 -0.103094
9.24642 8.22412 7.63714 -0.022114 0.001227 -0.012731
12.50777 12.80118 5.49172 0.022669 -0.022502 -0.021867
8.82055 13.36139 9.32827 0.000851 0.040048 0.009219
11.17945 6.63861 9.32827 -0.000851 -0.040048 0.009219
10.00000 10.00000 9.81639 0.000000 0.000000 -0.033824
6.27211 10.54758 5.49736 0.001459 -0.014796 0.009705
7.49223 7.19882 5.49172 -0.022669 0.022502 -0.021867
5.94690 13.32990 7.80975 -0.821818 0.493098 0.271541
11.05190 15.17530 7.80624 0.282254 0.967278 0.205919
14.05310 6.67010 7.80975 0.821818 -0.493098 0.271541
16.41560 8.84240 7.80915 0.484998 -0.673830 0.056259
5.62430 5.16080 7.73904 0.096844 -0.983436 0.135032
8.94810 4.82470 7.80624 -0.282254 -0.967278 0.205919
3.58440 11.15760 7.80915 -0.484998 0.673830 0.056259
14.37570 14.83920 7.73904 -0.096844 0.983436 0.135032
-----------------------------------------------------------------------------------
total drift: -0.000087 -0.000072 0.152962

The total drift indicates how accurate the calculations are. There are several contributions. I suggest you carefully read the corresponding section from the Vasp-workshop talks:

http://cms.mpi.univie.ac.at/vasp-worksh ... curacy.pdf

Especially have a look at the chart at page 28 in order to get an idea of things you can do to decrease the drift. The issue has also been discussed previously in the forum and if you search the threads you will probable find further discussions.

Best regards,
/Dan

<span class='smallblacktext'>[ Edited Tue Sep 02 2008, 09:21PM ]</span>

1) Certainly, the PREC=Low is the main reason of the drift.
2) If the drift remains after increasing PREC, please mind the following:
if a unit cell contains vaccum (like in a slab or cluster) it may me helfpful to fix at least one atom of the cell in its position to avoid a drift of the species through the vacuum. (you have to use selective dynamics to do so)