BUG: MAGMOM removal in (HSE)BandStructureMaker

[fraricci]

Hi there!

When using a (HSE)BandStructureMaker I noticed that the NSCF calculations (bands and dos) does not have the MAGMOM in the INCAR and I get the following warning in the OUTCAR:

|   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.

And results, i.e. the DOS, are quite different between the cases with and without keeping the MAGMOM.

Now, looking at the VASP wiki https://www.vasp.at/wiki/MAGMOM I see the following:
"When restarting a magnetic calculation, MAGMOM is only used to determine the symmetry of the system and not to set the on-site magnetic moment. Therefore, if you remove the MAGMOM tag before restarting from a converged WAVECAR or CHGCAR, the magnetization is likely to be symmetrized away."

So, I suppose that MAGMOM should remain in the INCAR of a NSCF calculations from the previous static one.

I think the lines that remove the MAGMOM are

atomate2/src/atomate2/vasp/sets/core.py

Line 262 in 6b19e55

updates["MAGMOM"] = None

atomate2/src/atomate2/vasp/sets/core.py

Line 523 in 6b19e55

updates["MAGMOM"] = None

What seems odd is that those lines seem to be added a long time ago and nobody complained.
Could this warning be added in recent versions of VASP? I'm using the 6.4.2.

Let me know what you think.

Thanks!

[esoteric-ephemera]

Thanks @fraricci, I think this maybe due to the ambiguity around setting MAGMOM in VASP - naively, I would assume that specifying CHGCAR or WAVECAR would also set MAGMOM, but that doesn't appear to be the case. Guessing this is also confusing for others

I'd be in favor of removing the MAGMOM updates - happy to take care of this

[fraricci]

Thanks @esoteric-ephemera, yes, that's what I thought too! At first I saw that the magmoms was not in the incar, I said, ok, it can read them from the CHGCAR or WAVECAR, but then I saw the warning in the outcar, then the results were clearly those of a FM system (which I did not want). Then, I saw the VASP wiki...

With @gpetretto, we looked at the vasp code and saw that the warning was added in VASP 6.2.1 or before, but after the 5.4.4. The Wiki was updated later on.

I did not check the VASP forum for further details; I could, if it helps.
I did search, but did not find anything related to this warning and restarting from previous calculation.
Also, I did not check what atomate was doing, I'd bet it keeps the magmoms.

My "worry" is how many wrong calculations have been made so far ;-) Hopefully, the user realized it.

Anyway, I did test commenting those lines and the NCSF INCAR contain the initial MAGMOM.
So, if you want to change that, please go ahead.

If someone else has feedback, please.

[JaGeo]

@fraricci that's a good hint in general. I haven't been running many magnetic calculations with atomate2 so far. It could be an issue in several of the workflows. I am aware we did some changes in the phonon workflow to make it more robust. Potentially, we should test or inspect this systematically...