I did static self-consistent calculation with IBRION=-1 after ionic relaxation, however, it wasn't convergent until 60 electronic iterations had performed. Even in the structural relaxation, there were 30~40 electronic iterations in each ionic relaxation. My INCAR is listed as follow and the system is MnO2. I wonder how to modify my parameter to reach convergence.
start parameter for this run :
PREC = accurate
ISTART = 0
ICHARG = 2
ISPIN = 2
Electron Relaxation
ALGO = Fast
LREAL = Auto
EDIFF = 1E-6
Ionic Relaxation
IBRION = -1
ENCUT = 600
NSW = 0
EDIFFG = -0.01
ISMEAR = 0
SIGMA = 0.05
LMAXMIX = 6
ISIF = 3
GGA = 91
VOSKOWN =1
NGX = 66
NGY = 66
NGZ = 66
MAGMOM =16*6 32*0
AMIX = 0.2
BMIX = 0.0001
ADDGRID = .TRUE. NELMDL=8
LDAU = .TRUE.
LDAUTYPE = 1
LDAUL = 2 0
LDAUU = 4.84 0
LDAUJ = 0 0 0
LDAUPRINT = 0
Write flag
LWAVE = .FALSE.
LCHARG = .FALSE.
LVTOT = .FALSE.
Convergence of electronic calculation
Moderators: Global Moderator, Moderator
-
tssunrise
- Newbie
- Posts: 6
- Joined: Fri Jul 09, 2010 7:13 am
Convergence of electronic calculation
Last edited by tssunrise on Tue Apr 26, 2011 12:23 am, edited 1 time in total.
-
boris
- Full Member
- Posts: 106
- Joined: Wed Jan 16, 2008 1:27 pm
- License Nr.: 567
- Location: France
Convergence of electronic calculation
Hi
You are using DFT+U and the convergence can be pretty long. You can also use LCHARG=.TRUE. and ICHARG=1 to start from the charge density calculated in the relaxation run.
Actually, with DFT+U, it is strongly recommended to use ICHARG=1 otherwise from one calculation to another, you will reach different metastable states.
Regards
You are using DFT+U and the convergence can be pretty long. You can also use LCHARG=.TRUE. and ICHARG=1 to start from the charge density calculated in the relaxation run.
Actually, with DFT+U, it is strongly recommended to use ICHARG=1 otherwise from one calculation to another, you will reach different metastable states.
Regards
Last edited by boris on Tue Apr 26, 2011 7:12 pm, edited 1 time in total.
-
sketrat
Convergence of electronic calculation
Hi,
I am facing the same problem to tssunrise's on the static calculation.
My calculation is DFT+U on CeO2 slab model. The calculation reaches 100 electronic iteration steps but still does not converge (even the charge density from converged optimization calculation, ICHARG=1, is restored). Do we have any other options to deal with this difficult convergence?
Thank you in advanced
best wishes,
sketrat
My INCAR is shown below
System = CeO2(111) PW91_U5 5.494
Starting parameters for this run:
NWRITE = 2
ISTART = 0
ICHARG = 1
INIWAV = 1
ISPIN = 2
Electronic Relaxation:
PREC = high
NELM = 100
NELMIN = 2
NELMDL =-5
ENCUT = 408
EDIFF = 1E-5
LREAL = .TRUE.
IALGO = 48
LCHARG = .TRUE.
LWAVE = .TRUE.
LELF = .TRUE.
LVTOT = .FALSE.
Ionic Relaxation:
NSW = 0
IBRION = -1
EDIFFG = -0.01
ISIF = 2
ISYM = 0
DOS related values:
LORBIT = .FALSE.
ISMEAR = 0
SIGMA = 0.1
NGX = 90
NGY = 90
NGZ = 256
AMIX = 0.10
AMIX_MAG = 1.00
BMIX = 0.10
BMIX_MAG = 0.50
LMAXMIX = 6
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 3 -1
LDAUU = 5 0
LDAUJ = 0 0
LDAUPRINT = 1
NPAR = 4
LPLANE = .TRUE.
LSCALU = .FALSE.
NSIM = 4
I am facing the same problem to tssunrise's on the static calculation.
My calculation is DFT+U on CeO2 slab model. The calculation reaches 100 electronic iteration steps but still does not converge (even the charge density from converged optimization calculation, ICHARG=1, is restored). Do we have any other options to deal with this difficult convergence?
Thank you in advanced
best wishes,
sketrat
My INCAR is shown below
System = CeO2(111) PW91_U5 5.494
Starting parameters for this run:
NWRITE = 2
ISTART = 0
ICHARG = 1
INIWAV = 1
ISPIN = 2
Electronic Relaxation:
PREC = high
NELM = 100
NELMIN = 2
NELMDL =-5
ENCUT = 408
EDIFF = 1E-5
LREAL = .TRUE.
IALGO = 48
LCHARG = .TRUE.
LWAVE = .TRUE.
LELF = .TRUE.
LVTOT = .FALSE.
Ionic Relaxation:
NSW = 0
IBRION = -1
EDIFFG = -0.01
ISIF = 2
ISYM = 0
DOS related values:
LORBIT = .FALSE.
ISMEAR = 0
SIGMA = 0.1
NGX = 90
NGY = 90
NGZ = 256
AMIX = 0.10
AMIX_MAG = 1.00
BMIX = 0.10
BMIX_MAG = 0.50
LMAXMIX = 6
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 3 -1
LDAUU = 5 0
LDAUJ = 0 0
LDAUPRINT = 1
NPAR = 4
LPLANE = .TRUE.
LSCALU = .FALSE.
NSIM = 4
Last edited by sketrat on Thu Apr 28, 2011 2:04 am, edited 1 time in total.