Page 1 of 1
CI-NEB
Posted: Wed Oct 30, 2024 4:59 pm
by msaad22@uic.edu
Hello,
I am trying to make CI-NEB for a molecule on zerovalent iron (97 atom), but
it looks like the calculation is very expensive (1 SCF takes 1 day). I
wonder if there is a way to make it less expensive.
K-points
3*3*1
Here is my INCAR paprameters
SYSTEM = 4 layers Fe
ISTART = 0 ; ICHARG=2
ENCUT = 400
ISMEAR = 1 ; SIGMA = 0.2
IBRION=2
ISIF=2
NELM = 500
NSW = 100
ISPIN = 2
MAGMOM = 122*2
ALGO = Fast
EDIFF = 1E-6
EDIFFG = -0.02
IVDW = 11
ISYM = 0
LREAL= Auto
NCORE = 8
IMAGES = 4
SPRING = -5
LCLIMB = .TRUE.
Re: CI-NEB
Posted: Thu Oct 31, 2024 9:37 am
by jonathan_lahnsteiner2
Dear msaad22,
You can always make a calculation cheaper by reducing the number of KPOINTS or the ENCUT parameter. But making a calculation cheaper is always paid with the price of accuracy. Maybe an option for you would be to test your calculation with a cheaper setup, for example with a smaller simulation box. After verification that the calculation works as expected and the ideal parameters were found you can go to the more expensive setup.
In your INCAR file you are setting the NCORE tag. If this tag is chosen unlucky the simulation time will drastically increase. You can check the vasp wiki about setting NCORE. Also the tags NPAR and KPAR will be helpful for speeding up your calculation.
All the best Jonathan
Re: CI-NEB
Posted: Tue Nov 05, 2024 2:41 am
by msaad22@uic.edu
Thanks Jonathan for the quick response. It is very helpful. Now I have another problem that after the convergence, all images either ended up with the initial state or the final state resulting in no saddle point. How can I solve this problem?
Thanks,
Re: CI-NEB
Posted: Thu Nov 07, 2024 8:29 am
by jonathan_lahnsteiner2
Dear msaad22,
The reason might be that the structures you are using are chosen unlucky. But without seeing the POSCAR files and your OUTCAR files it is impossible to tell for sure. Please upload your input files and output files according to the vasp forum guidelines. The POSCAR, INCAR, KPOINTS and if possible OUTCAR files would be very helpful. Maybe you can take a look at Nudged_elastic_bands or Collective jumps of a Pt adatom on fcc-Pt (001): Nudged Elastic Band Calculation in the vasp wiki. This might be of help as well.
All the best Jonathan