Hard to converge using an already converged WAVECAR

[zhangguangping]

Dear all,

I want to do a dipole correction in vasp for a metal slab system. It is hard to converge directly for this system. So, I first run a job without the dipole correction and get a converged WAVECAR (say WAVECAR1), then I use this converged WAVECAR1 to initalize the job with dipole correction and get a converged WAVECAR2 very easily. However, I found if I use the converged WAVECAR2 to initalize the same job with dipole correction, it could not converge at all. It was expected to converge much more easily since it was a already converged WAVECAR for this calculation. If this not a bug for VASP, then it would bring trouble for a geometric optimization of a slab system with a dipole correction.

With best regards,

Guangping Zhang

[admin]

please check if the position of the origin of the dipole moment remains constant or if it moves
(eg due to charge sloshing in the vacuum)

[zhangguangping]

please check if the position of the origin of the dipole moment remains constant or if it moves
(eg due to charge sloshing in the vacuum)

Thanks for your reply.

How can I check the position of origin of the dipole moment? Or can I fix the position of the origin in the calcualtion? How I guarantee this in a geometric optimization with dipole correction in a slab system, since the dipole correction is important for the force calcualtion, hence the geometric optimization.

With best regards,

Guangping