Page 1 of 1

Convergence problems with dipole correction

Posted: Tue Jun 21, 2022 6:29 pm
by weronika_walkosz3
Hello,
I am trying to model Rh(111) in the presence of oxygen adsorbed at one side only. Hence, I am using dipole corrections (LDIPOL = T, IDIPOL = 3). With increasing O coverage, I face energy convergence problems. Can someone comment on how to overcome this issue? Some of my jobs converged when dipole correction was not used.

Re: Convergence problems with dipole correction

Posted: Tue Jun 21, 2022 7:08 pm
by fabien_tran1
Hi,

Could you please provide the INCAR, POSCAR, KPOINTS and OUTCAR files.

Re: Convergence problems with dipole correction

Posted: Tue Jun 21, 2022 9:53 pm
by weronika_walkosz3
Thanks for the reply; I appreciate it.
How do I attach the files (is it possible) to the response? OUTCAR is too big to share otherwise.

Re: Convergence problems with dipole correction

Posted: Wed Jun 22, 2022 7:30 am
by fabien_tran1
The most convenient way is to create a zip file and then attach it by clicking on "Attachments". If the OUTCAR is too large, then just grep for TOTEN in OUTCAR ("grep TOTEN OUTCAR") and display us the results (I just want to see how the total energy behaves along the iterations).

Re: Convergence problems with dipole correction

Posted: Wed Jun 22, 2022 3:48 pm
by weronika_walkosz3
Attached are the files. Thanks so much in advance for your help.

Re: Convergence problems with dipole correction

Posted: Thu Jun 23, 2022 8:30 am
by fabien_tran1
Hi,

Your calculation with dipole correction stopped because of "Error EDDDAV: Call to ZHEGV failed ..." (last line in OUTCAR). At iteration 37, the density and orbitals were so strange that the diagonalization of the Hamiltonian failed. To some extent this may be a question of luck, since this error could have happen (or not happen) at any early stage of the calculation when the density/orbitals have still not acquired a reasonable shape. For the moment, a suggestion is to restart the calculation with a different mixing scheme (see IALGO) or changing AMIX or BMIX for instance.

By the way, did you restart the calculation with dipole correction using the WAVECAR from the calculation without dipole correction as suggested at LDIPOL (where the DIPOL tag is also discussed)?

There is also this related (but old) topic here: https://www.vasp.at/forum/viewtopic.php?f=4&t=13121

Re: Convergence problems with dipole correction

Posted: Mon Jun 27, 2022 6:52 pm
by weronika_walkosz3
Thanks so much.