VASP Forum

Dear Experts,
I am performing ab-initio molecular dynamics (AIMD) recoil simulations to estimate the threshold displacement energy (Td) in bulk NiO by considering a 3X3X3 supercell. A primary knock-on atom (Ni) is given an initial velocity corresponding to ~20 eV recoil energy along the [111] direction. I am not fully aware of these calculations and am trying first time. However, the calculation consistently stops after a few MD steps with the following error:

ERROR in subspace rotation PDSYEVX/PZHEEVX: I2,IFAIL = 1 364
----> I REFUSE TO CONTINUE WITH THIS SICK JOB ... BY

I am attaching my INCAR, KPOINTS, OUTCAR, POSCAR, and POTCAR files for your reference. I am not getting why this error is coming, and what can be done now. I would be grateful if you could please see my input files and help me to rectify this error.

https://drive.google.com/drive/u/1/fold ... MYvs-c4twO

Dear alok_shukla1,

I see that in your calculation the atoms are not moving from their equilibrium positions. The reason is that you start from the equilibrium configuration and run molecular dynamics in the NVE ensemble. Since all atoms are initially at their equilibrium positions, the forces are negligibly small, and therefore there is practically no atomic motion.

If I understand correctly, you are trying to run a recoil MD simulation and assign an initial velocity to one of the atoms. The issue here is the incorrect format of the POSCAR file. You need to specify the velocity mode and then provide velocities for all atoms, as explained on the wiki. In particular, you should remove the line Velocities.

You can also see in your OUTCAR file that the velocities were not read:
No initial velocities read in.

Best,
Alexey

Dear Expert,
After correcting my POSCAR formatting, my job started working nicely. I provided the initial velocity in the [111] direction in my first attempt. Now, I also tried it in the [100] direction as well. Although the [111] job is still running, the [100] job stops, and I am receiving the error. Please see the attached vasp.err file. I am also attaching my INCAR, KPOINTS, POTCAR, POSCAR, and OUTCAR files.
I hope this time I am doing everthing correct, but an unexpected error is received.
https://drive.google.com/file/d/1WYDQN1 ... sp=sharing

I see that the issue with this calculation is that the SCF doesn't converge in the NELM steps. The likely reason is that your time step is too large. I would suggest that you decrease POTIM and increase NELM.

Also, your OUTCAR is really large. You might want to use a less verbose output to reduce the size of OUTCAR, i.e., LDAUPRINT, NBLOCK, and KBLOCK.