I’m trying to do the finite electric field calculation (berryopt=14) for antiferroelectric materials (40 atoms, nkpt=10).
It takes very long time, so I would like to use kgb parallelisation (LOBPCG or Chebfi).
I set up the flags (paral_kgb=1, npband=2, npkpt=10, npfft=1) and ran the calculation, but the calculation didn’t work.
I would like to know whether the band parallelization is available for finite electric field calculation.
I found a similar question in forum (URL : [SOLVED] band parallelisation in finite electric field? - ABINIT Discussion Forums), but it was too old.
Could you teach me the recent information?
can you try parallelization with just the k-points please? Without the parallelization for bands and fft.
Dear Joao Abreu
Thank you for your suggestion.
Sorry for my lack of explanation.
I already tried prallelization with just the k-points and it worked.
However I could not set nprocs more than nkpt(=10) and the calculation took several weeks.
Isn’t the kgb parallelisation for electric field calculation available yet?
In that case, is there any other way to inclease nprocs?
I was checking the code and indeed the parallelization was only made for the k-points and spin.
I will contact the responsible for this part of the code to have more information.
Thank you for checking.
I’m sorry for the trouble.
I confirm that the parallelkgb has not been implemented for Berry phase and applied electric field calculations. It should be done but we are missing hands to do it… So this is pending until further notice.