MAGMOM needed if rerun from CHGCAR/WAVECAR or no?

[reynaldo.putra]

The question I have is as stated by the title, because when I did not set the MAGMOM for a rerun of a calculation (ICHARG = 1), it still gives me the warning:

Code: Select all

|   You use a magnetic or noncollinear calculation, but did not specify
|   the initial magnetic moment with the MAGMOM tag. Note that a
|   default of 1 will be used for all atoms. This ferromagnetic setup
|   may break the symmetry of the crystal, in particular it may rule
|   out finding an antiferromagnetic solution. Thence, we recommend
|   setting the initial magnetic moment manually or verifying carefully
|   that this magnetic setup is desired.

Even though it is mentioned that the spin information is written in the CHGCAR and WAVECAR files. I would like to clear the ambiguity surrounding this setting/information.

And a smaller question, I would like to confirm if the wavefunctions of the WAVECAR files indeed contain information of the spin polarization/magnetization?

Thank you,
Reynaldo

[christopher_sheldon1]

Dear Reynaldo Putra,

Thank you for your question.

MAGMOM is still required. If you load CHGCAR/ WAVECAR, then it will begin from there, but if no MAGMOM is specified, the initial magnetic moments will be set to ferromagnetic. If you started from an antiferromagnetic density, then you will gradually converge to a ferromagnetic one. This can also be used to start a magnetic calculation from a non-magnetic density. To ensure the same magnetic structure, keep the MAGMOM constant in each run.

Both WAVECAR and CHGCAR contain spin information. The CHGCAR contains the total charge density and the magnetization density (see https://vasp.at/wiki/CHGCAR#Magnetic_calculations for the difference between spin-polarised and noncollinear). It writes the magnetic symmetry (i.e. MAGMOM), but this is not read when the calculation is restarted. Instead, it is rebuilt every run using MAGMOM.

WAVECAR contains the spin-resolved set of orbitals, eigenvalues, and occupation, but does not contain any information about MAGMOM.

So you receive this warning message because MAGMOM has been set to ferromagnetic. It is best practice to explicitly define this yourself for each calculation and keep it consistently, unless you have a good reason to change it (like starting from a non-magnetic density to produce a ferromagnetic, antiferromagnetic, etc).

Does this answer your question?

Best wishes,

Chris

[reynaldo.putra]

Dear Chris,

Thank you very much for the clarification.
So, what I am concerned about is, when I do an ionic relaxation -> SCF -> nSCF workflow to get the DOS, is it acceptable to simply use the same MAGMOM tags throughout? I imagine it would be correct instead if I checked the OUTCAR and modify the MAGMOM based on that to obtain physically correct results.

And one more question, when I use ICHARG = 1 to read the CHGCAR for the SCF calculation, does the CHGCAR file get updated after the SCF calculation?

Thank you again!
Best regards,
Reynaldo

[christopher_sheldon1]

Dear Reynaldo,

Yes, you should use the same MAGMOM in all three steps. MAGMOM only defines the initial magnetic moments when there is no density present. For subsequent steps, it sets the magnetic symmetry and not the magnetic moments. Sorry that I didn't make that clearer in my first reply. You could change the MAGMOM based on the OUTCAR but so long as the magnetic symmetry is the same, it will still set up the same calculation.

And yes, when you use ICHARG = 1, the old CHGCAR is read at the start of the calculation, then written over with the new density at the end. If you want to keep the old density, it's important to take a copy of the density before the calculation. If you want to keep the density the same and not overwrite it, e.g., for DOS, then you should use ICHARG = 11.

Best wishes,
Chris