Dipole correction and forces

Message

Vlad · #1 Post by **Vlad** » Tue Aug 08, 2006 5:36 pm

I am optimizing geometry of a charged cluster. I have IDIPOL=4 and LDIPOL=.TRUE. (cubic cell)

I do not understand how to interpret the forces in the OUTCAR file. 6 coordinates are fixed, so there should be no drift with forces this low. The forces also do not seem to add up to the drift:

POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
6.00088 7.82676 6.38200 0.001898 -0.002571 -0.009659
7.58330 5.08164 6.38138 -0.003856 -0.001857 -0.003847
4.41311 5.08733 6.38355 0.002245 0.002342 -0.002705
6.00000 6.00000 8.99580 -0.005684 0.007369 0.007092
7.78333 7.02091 5.58974 -0.003839 0.001507 -0.007364
4.21629 7.02649 5.59270 0.002135 -0.003528 -0.007692
6.00000 8.04826 8.52747 0.004298 -0.002899 0.008740
5.99710 3.94284 5.59276 0.002948 0.003571 -0.008764
7.77519 4.97192 8.52305 -0.002150 0.006219 0.006558
4.22170 4.97924 8.52566 0.004537 0.004880 0.007349
6.00000 6.00000 7.04534 -0.002532 -0.015033 0.010293
-----------------------------------------------------------------------------------
total drift: -0.015660 -0.029627 7.952044

FORCES: max atom, RMS 0.010174 0.008090
Quasi-Newton relaxation of ions (Broydens 2nd method)
g(Force) = 0.347E-04 g(Stress)= 0.000E+00

Can anyone explain how to interpret these numbers?

Thank you in advance-
Vladimir

#2 Post by **admin** » Thu Aug 10, 2006 11:07 am

a considerable total drift of the forces may indicate that there are significant aliasing errors. Please increase the PREC and set ENCUT to at least 1.3*(max(ENMAX)) of POTCAR. It may also help to use harder pseudopotentials if they are available for the elements of your cluster.