"reached required accuracy" without meeting the required EDIFFG

Message

cenwanglai · #1 Post by **cenwanglai** » Sat May 08, 2010 3:24 am

INCAR:
PREC=Accurate
ENCUT=400
EDIFF=1E-4
EDIFFG=-0.05
IBRION=2
NSW=10
ISMEAR=0
SIGMA=0.05

POSCAR with selective dynamic type,three bottom atomic layers were fixed by "F F F",others were "T T T".

KPOINTS for slab with 13 Angstrom vacuum layer.
Auto
0
M-P
1 1 1
0 0 0

POTCAR paw-gga

This vasp work stoped in less than NSW ion steps and shew "reached required accuracy......." in the OUTCAR.For each electronic steps, the EDIFF criterion is OK.

BUT,the total drift read like: -0.000003 -0.000005 0.033633
AND, TOTAL-FORCEs of a lot of atoms are beyond the EDIFFG criterion.

Can you tell me what is the problem?

jzzhao · #2 Post by **jzzhao** » Sat May 08, 2010 11:49 am

you have fixed some atoms, there force were not considered for convergence criterion.

cenwanglai · #3 Post by **cenwanglai** » Sat May 22, 2010 1:56 am

Yes, I have got it. Thank you for your repy.

VASP001 · #4 Post by **VASP001** » Sat May 22, 2010 11:36 pm

Maybe you can try like this
mv CONTCAR to POSCAR ,and set IBRION =1 ,ADDGRID=.TRUE.