structural optimization becomes unstable
Posted: Sun Jun 08, 2014 9:45 pm
I tried to do structural optimization with BaWO4,
but for some reason the procedure doesn't converge. Here is 'grep tolmxf log':
Initially, the convergence seems promising, but then it worsens until the calculation stops with an error. What could cause such behavior? I tried changing several input paramaters and different PAW datasets but nothing helps. Also, I should mention that there are no problems with CaWO4, for which I use an almost identical input. I have attached the output files.
Code: Select all
# BaWO4
ndtset 2
###################
# Crystal structure
###################
scalecart 2*5.6134 12.720 angstrom
rprim
-0.5 0.5 0.5
0.5 -0.5 0.5
0.5 0.5 -0.5
ntypat 3
znucl 56 74 8
natom 12
typat 1 2 3 3 1 2 3 3 3 3 3 3
xred 5.0000000000E-01 5.0000000000E-01 0.0000000000E+00
0.0000000000E+00 0.0000000000E+00 0.0000000000E+00
2.0080000000E-01 3.1130000000E-01 3.5590000000E-01
9.5540000000E-01 8.4490000000E-01 6.4410000000E-01
2.5000000000E-01 7.5000000000E-01 5.0000000000E-01
7.5000000000E-01 2.5000000000E-01 5.0000000000E-01
6.1300000000E-02 2.0540000000E-01 6.1050000000E-01
5.9490000000E-01 4.5080000000E-01 3.8950000000E-01
5.4920000000E-01 9.3870000000E-01 1.4410000000E-01
7.9460000000E-01 4.0510000000E-01 8.5590000000E-01
6.8870000000E-01 4.4600000000E-02 8.8950000000E-01
1.5510000000E-01 7.9920000000E-01 1.1050000000E-01
###############
# BZS and bands
###############
ngkpt 3*4
shiftk 3*0.5
nbdbuf 5
nband 53
####################
# Molecular dynamics
####################
ionmov 3
ecutsm2 0.5
optcell1 0
optcell2 2
#################
# Cutoff energies
#################
ecut 22
pawecutdg 60
############
# Tolerances
############
toldff 1.0d-6
tolmxf 1.0d-5
dilatmx2 1.05
nstep 50
ntime 50
###############
# Data handling
###############
prtwf 0
prtden 0
getxred -1
######
# Misc
######
diemac 4
but for some reason the procedure doesn't converge. Here is 'grep tolmxf log':
Code: Select all
max grad (force/stress) = 8.7684E-03 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 3.7869E-03 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 4.7618E-04 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 4.9138E-04 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 9.3201E-03 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 9.5312E-02 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 2.6146E-01 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 7.8148E-01 > tolmxf= 1.0000E-05 ha/bohr (free atoms)
max grad (force/stress) = 1.5120E+00 > tolmxf= 1.0000E-05 ha/bohr (free atoms)Initially, the convergence seems promising, but then it worsens until the calculation stops with an error. What could cause such behavior? I tried changing several input paramaters and different PAW datasets but nothing helps. Also, I should mention that there are no problems with CaWO4, for which I use an almost identical input. I have attached the output files.