[CP2K-user] [CP2K:19953] Re: Improving convergence for bulk Iron (bcc Fe)

[fsai...@gmail.com]

Hello,
Have you tried to replicate the unit cell, say 5x5x5, and check if the SCF 
cycles converge with this large system ?
Regards.
 - Fernan

On Monday, January 30, 2023 at 12:01:51 AM UTC+1 Бранислав Миловановић 
wrote:

> Dear CP2K users,
>
> I'm struggling with convergence for bulk Fe, i.e. 3x3x3 supercell 
> containing 54 Fe atoms. I wanted to simulate adsorption of different 
> organic molecules on Fe(100) slab later, so this is my starting point.
>
> During GEO_OPT or CELL_OPT runs I sporadically encounter SCF issues like 
> this:
> ...
>     33 Broy./Diag. 0.70E+00    2.7     1.02510462     -7902.0344477276 
>  2.42E+01
>     34 Broy./Diag. 0.70E+00    2.7     1.02694146     -7882.2903511652 
>  1.97E+01
>     35 Broy./Diag. 0.70E+00    2.7     1.02572163     -7833.4164745080 
>  4.89E+01
>     36 Broy./Diag. 0.70E+00    2.7     1.02859994     -7878.1467068004 
> -4.47E+01
>     37 Broy./Diag. 0.70E+00    2.7     1.02924599     -7812.2513589335 
>  6.59E+01
> ...
>
> and here is my DFT section:
>
>  &DFT
>     BASIS_SET_FILE_NAME ${DIR}/BASIS_MOLOPT
>     POTENTIAL_FILE_NAME ${DIR}/GTH_POTENTIALS
>     UKS .TRUE.
>     MULTIPLICITY 123
>     RELAX_MULTIPLICITY 0.1
>     CHARGE 0
>     EXCITATIONS NONE
>     PLUS_U_METHOD MULLIKEN          # LOWDIN, MULLIKEN_CHARGES
>     &SCF
>       MAX_SCF 300
>       EPS_SCF 1.0E-6
>       SCF_GUESS RESTART
>       ADDED_MOS 200
>       NOTCONV_STOPALL .FALSE.
>       CHOLESKY RESTORE
>       &DIAGONALIZATION
>         ALGORITHM  STANDARD
>         EPS_ADAPT 0.0                 # Default 0.0
>         MAX_ITER 2                    # Default 2
>       &END DIAGONALIZATION
>       &MIXING
>         METHOD BROYDEN_MIXING
>         ALPHA 0.1                    # Default 0.4
>         BETA 1.2                      # Default 0.5
>         NBUFFER 6
>          NMIXING 2
>         NSKIP 0
>       &END MIXING
>       &SMEAR
>         ELECTRONIC_TEMPERATURE [K] 500
>         METHOD FERMI_DIRAC
>       &END SMEAR
>       &PRINT
>         &RESTART
>           BACKUP_COPIES 0
>         &END RESTART
>       &END PRINT
>     &END SCF
>     &MGRID
>       NGRIDS 4
>       CUTOFF 500
>       REL_CUTOFF 50
>     &END MGRID
>     &QS
>       METHOD GPW
>       EPS_DEFAULT 1.0E-12           # Default 1.0E-10
>       EXTRAPOLATION  ASPC           # Default
>       EXTRAPOLATION_ORDER 3         # Default
>     &END QS
>     &XC
>       &XC_FUNCTIONAL PBE
>       &END XC_FUNCTIONAL
>       &VDW_POTENTIAL
>         POTENTIAL_TYPE  PAIR_POTENTIAL
>         &PAIR_POTENTIAL
>           R_CUTOFF 1.0583544171800002E+01         # Default
>           TYPE DFTD3
>           PARAMETER_FILE_NAME ${DIR}/dftd3.dat
>           REFERENCE_FUNCTIONAL PBE
>         &END PAIR_POTENTIAL
>       &END VDW_POTENTIAL
> #      &XC_GRID
> #        XC_SMOOTH_RHO NN10
> #        XC_DERIV SPLINE2_SMOOTH
> ##        USE_FINER_GRID F
> #      &END XC_GRID
>     &END XC
>   &END DFT
>
> What are your thoughts about parameters that I use? Do I set 
> MAGNETIZATION, MULTIPLICITY and RELAX_MULTILIPLICITY correctly?
>
> I followed the instructions from the previous discussions here in group 
> and assembled the attached input file. Also, I looked for discussions 
> regarding metallic systems including Fe such as:
> https://groups.google.com/g/cp2k/c/ugkJLz4PSSI/m/C-ProBR6BwAJ
>
> I have an experience with the CP2K and molecular systems but not for solid 
> state calculations.
> Also, I have a couple of questions:
> 1) How does CP2K copes with SYMMETRY set to .TRUE. Is it better to leave 
> system to break symmetry?
> 2) Is it good to use XC_SMOOTH_RHO to smooth XC grid in this case?
>
> Thanks in advance,
> Branislav
>  
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/e6ea04c5-8e81-4d34-8268-8dbe779eab07n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240222/ee24698e/attachment.htm>