I have been running a job by VASP. My structure is a cluster(46 atoms) + one metal. when i want to optimize it, at the middle of the run, i saw this message: please rerun with smaller EDIFF, or copy CONTCAR
to POSCAR and continue ....and it doesn't converge...
my INCAR file is:
PREC = Normal
EDIFF = 1e-5
EDIFFG= -1e-3
ENCUT = 400
NSW = 3000
!ISPIN = 2
IBRION = 2
ISMEAR = 0 ; SIGMA = 0.0050
IVDW = 11
LORBIT = 11
NELM = 120
Hi Somayeh,
Your EDIFF value is 10 times stricter than the default. That makes it much harder to reach convergence.
Your EDIFFG value is also very strict, many people use something like -0.01 for relaxations.
Your smearing width is several dozen times smaller than the default. That also doesn't help convergence.
I suggest to first retry with values closer to the defaults for these three settings.
Kind regards,
Peter
Thank you Peter for your answer. I think you're right. Actually i try to rerun according what you said.
Best regards
Peter Klaver
Hi Somayeh Sayahi,
Your problem is the atomic structural convergence is hard.
You can try to increase the accurancy of SCF (smaller EDIFF) to get more accurate electron density in every relaxation cycle.
At the same time, you can also try to use IBRION = 3.
I hope this will helps.
When optimizing a cluster you must fix the coordinate of one atom (try with one close to the center of mass of your cluster) to avoid convergence problems, and better try ISIF=2.
Thank you dear for your answer. I will consider your advice
📷 Ana E. Torres
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Quantum Physics
- Quantum Mechanics