Calculations are not converging after Electric field is applied

[nirmal_ganguli2]

Facing convergence problem for slab calculation as soon as Electric field is applied for non-collinear calculation. There is no error after running the calculation, even if running it for a more extended time. Even after these many iterations
DAV: 290 -0.595534535353E+03 -0.84070E-03 -0.68698E-04144448 0.567E-02 0.579E-01
DAV: 291 -0.595534049647E+03 0.48571E-03 -0.13563E-04174816 0.569E-02 0.581E-01
DAV: 292 -0.595534117062E+03 -0.67415E-04 -0.57259E-05152912 0.340E-02 0.577E-01
DAV: 293 -0.595534505197E+03 -0.38814E-03 -0.17372E-05158856 0.186E-02 0.579E-01
DAV: 294 -0.595535565415E+03 -0.10602E-02 -0.22035E-04176760 0.707E-02 0.585E-01
job is getting killed because of the time constraint, but not converging. I have tried running the calculations with different versions of Vasp. I recently performed it with the 6.4.1 version, but that could not solve the issue.

[jonathan_lahnsteiner2]

Dear nirmal_ganguli2,

You could set a higher time limit in your job script. With the slurm job scheduler you could do this with:

Code: Select all

#SBATCH --time=72:00:00

This would set the time limit to 72 hours. Another thing you could try is switching
to a different electronic minimization algorithm. For this I would like to refer you to the IALGO and ALGO tag in the wiki
wiki/index.php/IALGO
wiki/index.php/ALGO

I hope this resolves your issue.

All the best Jonathan

[nirmal_ganguli2]

Dear Jonathan,
I tried IALGO and ALGO tags with a time limit of 72 hours, but the calculation is still not converged.

WARNING in EDDRMM: call to ZHEGV failed, returncode = 8 4 87
RMM: 69 -0.597546257190E+03 -0.47142E-03 -0.74852E-04239306 0.346E-02 0.287E-01
WARNING in EDDRMM: call to ZHEGV failed, returncode = 6 3 58
WARNING in EDDRMM: call to ZHEGV failed, returncode = 6 3 60
this I'm getting with IALGO = 48

and with IALGO = 38, nothing converged even after 130 iterations.

[jonathan_lahnsteiner2]

Dear nirmal_ganguli2,

You could check if your structure is optimized properly in the case when the electric field is applied.
Non converging electronic cycles can be caused by non-equilibrated POSCAR files.

Or you could try to apply a lower electric field and check if the calculation would converge then. Like this you could go step by step to the field
which is desired.
Maybe this wiki page could be of help for you: wiki/index.php/LSORBIT
Because you could start your calculation from a collinear one and go to the non collinear one. If you try to do so, this forum entry could also be of interest for you:forum/viewtopic.php?f=4&t=19069

All the best Jonathan