Job pausing after certain time

[tanay11]

Hi,

While running vasp (6.3.1) on HPCs, I encountered an unexpected issue with my job. The output files are not getting updated after running for 1-2 days, even though the 'Time_Left' continues to decrease.

This is the second time I've experienced this problem. Initially, I thought it might be due to an issue with the login node or a connectivity problem, so I resubmitted the job. However, I am facing the same issue again. I even tried a different HPC cluster, but the problem persists. Even if the job has converged, the output should indicate that "the required energy has converged," but the OUTCAR file indicates that the job is stuck at a certain step.

This made me wonder whether this is the problem with the software or if the issue is something else.
I hope this clarifies my problem.

Thanks and Regards

[alexey.tal]

Dear tanay11,

Thank you for your question.

The issue here is that the circulation doesn't converge.

There is a number of things that make this system rather difficult to converge, i.e., slab with dipole corrections, spin-polarization, CO2 molecule on the surface which requires vdW interactions, and finally a very strong electric field.
I don't know if you have tried it already, but usually to achieve convergence in such a complex case one needs to gradually increase the "complexity".
I would suggest that before you do the full calculation you try to make sure that you can converge just a slab (no molecule and no field). Then I would try to introduce the molecule. And only after that I would add the field. Also, the EFIELD value in your OUTCAR appears to be really large.

Also, I noticed that you did not specify DIPOL, which can be used to improve the convergence with dipole corrections. See our wiki page on how to use the dipole corrections in a slab calculation.

Best wishes,
Alexey

[tanay11]

Hi Alexey,

Thank you for your response.

I apologize for not providing the background of my work earlier. I am currently evaluating the effect of electric fields on CO2 desorption from alloys. My goal is to determine the optimal value of the electric field that will effectively desorb CO2, as well as to identify higher values that might also be effective.

Additionally, the initial structure I provided is the relaxed structure following CO2 adsorption on the surface. To clarify, I first relaxed the bulk material, then performed surface relaxation, and subsequently conducted CO2 adsorption on the surface. Now, I am analyzing that structure in the presence of an electric field.
I would love to hear any suggestions you might have based on this information!

Thanks

[alexey.tal]

Thank you for the clarification.

I see that you have done the intermediate steps. So if I understood you correctly you were able to get the calculations converge with the molecule on the surface but without the applied field.
But I wonder now if you were able to achieve the convergence with a smaller value of the EFIELD?

[tanay11]

Hi Alexey,

Yes, I was able to converge the calculation with the molecule on the surface but without the applied field.
I haven't tried using a smaller electric field yet with this material. I mean, the jobs are in the queue.