Failed to achieve convergence when electric field is switched on
Posted: Tue Feb 09, 2010 2:06 am
Hi, I am trying to study the effect of electric field on the adsorption of gas atom on surface. without electric field , relaxation goes fine.However, I have problem to achieve relaxation convergence when i switch on the electric field in Z direction.
At first, the electronic relaxation can not achieve the accuracy within default 60 cycles. I have tried to increase it to 80.
I have also tried to change the mix parameter as
However, i still encounter problems.
For example, in OUTCAR, I find for one electronic relaxation step,
DIPCOR: dipole corrections for dipol
direction 3 min pos 6,
dipolmoment 0.000000 0.000000 0.912249 electrons x Angstroem
Tr[quadrupol] -2327.486431
energy correction for charged system 0.000000 eV
dipol+quadrupol energy correction -0.044904 eV
added-field ion interaction -5.346719 eV (added to PSCEN)
Broyden mixing:
rms(total) = 0.82333E-01 rms(broyden)= 0.82333E-01
rms(prec ) = 0.87869E-01
weight for this iteration 100.00
eigenvalues of (default mixing * dielectric matrix)
average eigenvalue GAMMA= 1.4383
5.3317 5.3317 4.9586 4.1630 2.3620 2.3620 2.3647 2.0575 1.3137 1.3137
0.7937 0.7937 1.1182 1.1182 0.9931 0.9931 0.8856 0.7228 0.7228 0.2608
0.2608 0.4432 0.3551 0.3551 0.1967 0.1335 0.0007 0.0026 0.0024
Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 54.04659881
Ewald energy TEWEN = 23745.15327744
-1/2 Hartree DENC = -28851.42252664
-V(xc)+E(xc) XCENC = 535.03161939
PAW double counting = 3169.92096210 -3221.15885353
entropy T*S EENTRO = 0.00000000
eigenvalues EBANDS = -1759.06692635
atomic energy EATOM = 6026.66363728
---------------------------------------------------
free energy TOTEN = -300.83221149 eV
energy without entropy = -300.83221149 energy(sigma->0) = -300.83221149
...
In the following step,
DIPCOR: dipole corrections for dipol
direction 3 min pos 144,
dipolmoment 0.000000 0.000000 -646.079567 electrons x Angstroem
Tr[quadrupol] -5173.133808
energy correction for charged system 0.000000 eV
dipol+quadrupol energy correction -22523.099855 eV
added-field ion interaction 68185.056883 eV (added to PSCEN)
Broyden mixing:
rms(total) = 0.12641E+02 rms(broyden)= 0.12638E+02
rms(prec ) = 0.16807E+02
weight for this iteration 100.00
eigenvalues of (default mixing * dielectric matrix)
average eigenvalue GAMMA= 1.3908
5.3376 5.3376 4.9702 4.1510 2.3617 2.3617 2.3875 2.0355 1.3145 1.3145
0.7937 0.7937 1.1188 1.1188 0.9929 0.9929 0.8858 0.7228 0.7228 0.2608
0.2608 0.4432 0.3551 0.3551 0.1967 0.1335 0.0000 0.0007 0.0026 0.0024
Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 45721.39525096
Ewald energy TEWEN = 23745.15327744
-1/2 Hartree DENC = -28850.43322811
-V(xc)+E(xc) XCENC = 535.04012131
PAW double counting = 3168.94090404 -3220.16189141
entropy T*S EENTRO = -0.00000016
eigenvalues EBANDS = -86861.30415609
atomic energy EATOM = 6026.66363728
---------------------------------------------------
free energy TOTEN = -39734.70608474 eV
energy without entropy = -39734.70608458 energy(sigma->0) = -39734.70608466
I am wondering, how can the dipole correction change so dramatically from one electronic step to the next one. I noticed the change of the line
from : direction 3 min pos 6,
to : direction 3 min pos 144,
what does 6 and 144 mean?
I am using INCAR like :
EDIFF=1E-5 # default is 0.0001, default EDIFFG is 10*EDIFF
NSW=1000
IBRION=2
ISIF=1 #no cell shape and no cell volumn, no tensor calculation, just force
PREC=ACU
ISMEAR=0
SIGMA=0.002
ISPIN=2
LORBIT=11
EFIELD=0.3
LDIPOL=.TRUE.
IDIPOL=3
AMIX = 0.2
BMIX = 0.0001 ! almost zero, but 0 will crash some versions
AMIX_MAG = 0.8
BMIX_MAG = 0.0001 ! almost zero, but 0 will crash some versions
NELM=70
IALGO = 38 algorithm
ISYM=2 # default for PAW
Also, My KPOINT is :
Automatic mesh
0
Monkhorst-Pack
8 8 2
0 0 0
Any suggestions to improve the convergence?
thanks
At first, the electronic relaxation can not achieve the accuracy within default 60 cycles. I have tried to increase it to 80.
I have also tried to change the mix parameter as
However, i still encounter problems.
For example, in OUTCAR, I find for one electronic relaxation step,
DIPCOR: dipole corrections for dipol
direction 3 min pos 6,
dipolmoment 0.000000 0.000000 0.912249 electrons x Angstroem
Tr[quadrupol] -2327.486431
energy correction for charged system 0.000000 eV
dipol+quadrupol energy correction -0.044904 eV
added-field ion interaction -5.346719 eV (added to PSCEN)
Broyden mixing:
rms(total) = 0.82333E-01 rms(broyden)= 0.82333E-01
rms(prec ) = 0.87869E-01
weight for this iteration 100.00
eigenvalues of (default mixing * dielectric matrix)
average eigenvalue GAMMA= 1.4383
5.3317 5.3317 4.9586 4.1630 2.3620 2.3620 2.3647 2.0575 1.3137 1.3137
0.7937 0.7937 1.1182 1.1182 0.9931 0.9931 0.8856 0.7228 0.7228 0.2608
0.2608 0.4432 0.3551 0.3551 0.1967 0.1335 0.0007 0.0026 0.0024
Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 54.04659881
Ewald energy TEWEN = 23745.15327744
-1/2 Hartree DENC = -28851.42252664
-V(xc)+E(xc) XCENC = 535.03161939
PAW double counting = 3169.92096210 -3221.15885353
entropy T*S EENTRO = 0.00000000
eigenvalues EBANDS = -1759.06692635
atomic energy EATOM = 6026.66363728
---------------------------------------------------
free energy TOTEN = -300.83221149 eV
energy without entropy = -300.83221149 energy(sigma->0) = -300.83221149
...
In the following step,
DIPCOR: dipole corrections for dipol
direction 3 min pos 144,
dipolmoment 0.000000 0.000000 -646.079567 electrons x Angstroem
Tr[quadrupol] -5173.133808
energy correction for charged system 0.000000 eV
dipol+quadrupol energy correction -22523.099855 eV
added-field ion interaction 68185.056883 eV (added to PSCEN)
Broyden mixing:
rms(total) = 0.12641E+02 rms(broyden)= 0.12638E+02
rms(prec ) = 0.16807E+02
weight for this iteration 100.00
eigenvalues of (default mixing * dielectric matrix)
average eigenvalue GAMMA= 1.3908
5.3376 5.3376 4.9702 4.1510 2.3617 2.3617 2.3875 2.0355 1.3145 1.3145
0.7937 0.7937 1.1188 1.1188 0.9929 0.9929 0.8858 0.7228 0.7228 0.2608
0.2608 0.4432 0.3551 0.3551 0.1967 0.1335 0.0000 0.0007 0.0026 0.0024
Free energy of the ion-electron system (eV)
---------------------------------------------------
alpha Z PSCENC = 45721.39525096
Ewald energy TEWEN = 23745.15327744
-1/2 Hartree DENC = -28850.43322811
-V(xc)+E(xc) XCENC = 535.04012131
PAW double counting = 3168.94090404 -3220.16189141
entropy T*S EENTRO = -0.00000016
eigenvalues EBANDS = -86861.30415609
atomic energy EATOM = 6026.66363728
---------------------------------------------------
free energy TOTEN = -39734.70608474 eV
energy without entropy = -39734.70608458 energy(sigma->0) = -39734.70608466
I am wondering, how can the dipole correction change so dramatically from one electronic step to the next one. I noticed the change of the line
from : direction 3 min pos 6,
to : direction 3 min pos 144,
what does 6 and 144 mean?
I am using INCAR like :
EDIFF=1E-5 # default is 0.0001, default EDIFFG is 10*EDIFF
NSW=1000
IBRION=2
ISIF=1 #no cell shape and no cell volumn, no tensor calculation, just force
PREC=ACU
ISMEAR=0
SIGMA=0.002
ISPIN=2
LORBIT=11
EFIELD=0.3
LDIPOL=.TRUE.
IDIPOL=3
AMIX = 0.2
BMIX = 0.0001 ! almost zero, but 0 will crash some versions
AMIX_MAG = 0.8
BMIX_MAG = 0.0001 ! almost zero, but 0 will crash some versions
NELM=70
IALGO = 38 algorithm
ISYM=2 # default for PAW
Also, My KPOINT is :
Automatic mesh
0
Monkhorst-Pack
8 8 2
0 0 0
Any suggestions to improve the convergence?
thanks