Total energy

[IBRAHIM]

Dear All,

For some surface slabs of transition metals: if we change number of nodes, vasp version, or even sometimes start from converged contcar (without wavecar), one can get different converged total energies in range of 1 eV (of course with different magnetic configuration).

Is it system related or code related?

Best regards,
Ibrahim

[alex]

Hi Ibrahim,

with the number of cores you might change NBANDS for performance reasons automatically. Check if this is all the same. VASP adds empty bands to allow equal distribution over all processes. This, however, does not affect the total energy that much. What you describe is probably due to the initial guess of the wavefunction, which is random. So you end up in different electronic configurations...

Cheers,

alex

[IBRAHIM]

Thank you very much Prof. Alex for your explanation.

In most cases there was no change in NBANDS, however, as you already mentioned that does not affect the total energy so much.
Also I agree with you that can be due to the initial guess of the wavefunction and results in different magnetic configurations.

Hopefully I can find away to stabilize those systems.

Best regards,
Ibrahim

[alex]

Hello Ibrahim,

there are two ways to force the spin into the desired configuration:
NUPDOWN and MAGMOM. Check the manual for both. You may also use them in combination.

Cheers,

alex

[IBRAHIM]

Thanks Prof. Alex for your recommendation.

Ibrahim