Page 1 of 1

relax problem

Posted: Fri Feb 08, 2008 3:38 am
by rudra-b
dera friends, i am trying to relax a Mn system, but inspite of several days trial, its not relaxing(the force is not going down to EDIFF). my INCAR is as follws:
SYSTEM = Mn13_CUB

ISTART = 1
# INIWAV = 0
# ICHARG = 0

NBANDS = 60

EDIFF = 0.001
EDIFFG = -0.01

IBRION = 1
NSW = 50
POTIM = 0.5
# NFREE = 10
PREC = medium

ISMEAR = -1
SIGMA = 0.001

#-------- For Spin Polarized Calculation --------
ISPIN = 2
NUPDOWN = 0
VOSKOWN = 1
#--------------------------------------------------
ALGO = F
LDIAG = .TRUE.
NELM = 40
NELMIN = 6
NELMDL = 0

# IMIX = 4
# AMIX = 0.2
# BMIX = 0.0001
# MAXMIX = 80
#--------------- save memory--------------

ISYM = 0
LCHARG = .FALSE.
LELF = .FALSE.
LWAVE = .TRUE.

#---------------------------------------


i tried out the different MIXING and POTIM, with no help
can anyone suggest something

relax problem

Posted: Fri Feb 08, 2008 10:16 am
by admin
1) are you sure that your system of 13 Mn atoms has zero net spin?
2) please check if each of the ionic steps are electronically converged properly after 40 steps (otherwise the forces will be unreliable).
3) EDIFF = 1e-03 may not be sufficient for electronic convergence,
therefore I suggest to use the default parameters at least (!, EDIFF=1e-04, NELM=60) if you have convergence problems.
4) maybe it helps to use MP (ISMEAR=1) instead of Fermi smearing (ISMEAR=-1) and increase SIGMA