[Wien] Total (MT+interstitial) up/dn charge density with SOC?

[Fecher, Gerhard]

Hall Lukasz,
some years ago, Jürgen Kübler told me the following about calculations of the properties for magnetic systems:
        If you think you have enough k-points, just use more !
therefore I use sometimes 256000 or 512000 k-points.... and maybe just more.

I don't know whether he mentioned it in his textbook, but the easiest test is: Try to calculate the magnetocrystalline anisotropy of Fe and Ni.

By the way, if you are still interested in the magnetic symmetry, use https://iso.byu.edu/findsym.php

Ciao
Gerhard

DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy:
"I think the problem, to be quite honest with you,
is that you have never actually known what the question is."

====================================
Dr. Gerhard H. Fecher
Institut of Physics
Johannes Gutenberg - University
55099 Mainz
________________________________________
Von: Wien [wien-bounces at zeus.theochem.tuwien.ac.at] im Auftrag von Lukasz Plucinski [lukasz.galaxy.s3 at gmail.com]
Gesendet: Donnerstag, 12. März 2026 16:19
An: A Mailing list for WIEN2k users
Betreff: Re: [Wien] Total (MT+interstitial) up/dn charge density with SOC?

Dear Prof. Blaha,

Thank you, this clarifies the issue of the symmetries.

There is a remaining issue with the number of kpoints in the SCF.

I converge CrSb with SOC, rkmax=9, with ec E-7 and cc E-6 (I also need to use EMAX=20 in init_so_lapw, otherwise I am getting segmentation fault in lapwso). This example is with M100 (but with M001 it is similar). Here is the dependence of the total magnetization in a converged case.scf on the number of kpoints (this is the kpoint number setting from kgen in init_so_lapw, of course the true amount of k-points in the wedge is smaller):

2000 kpoints
:MMTOT:  SPIN MAGNETIC MOMENT IN CELL     =   -0.00220
20000 kpoints
:MMTOT:  SPIN MAGNETIC MOMENT IN CELL     =   -0.00083
100000 kpoints
:MMTOT:  SPIN MAGNETIC MOMENT IN CELL     =   -0.00016
200000 kpoints
:MMTOT:  SPIN MAGNETIC MOMENT IN CELL     =   -0.00008

I think the total moment should be zero. With large remaining total moment one gets spurious spin polarizations along directions where spin polarization should vanish due to symmetry. With 200000 kpoints it is already quite OK, but this number of kpoints seems an overkill. I do not understand why this convergence should depend so much on the number of k-points.

It could be something like searching for some anticrossings where spin-polarization (and Berry curv. etc) gets high, but I don't see this kind of problem in magnetic convergence of CrSb.

Best,
Lukasz


On Fri, Feb 27, 2026 at 4:14 PM Peter Blaha <peter.blaha at tuwien.ac.at<mailto:peter.blaha at tuwien.ac.at>> wrote:
I attach the relevant part of case.outsymso file.

You can see the classification of symops into
A (preserves M)
B (symop * time inv  preserve M)
C does not preserve M

All 12 operations belong to A or B.

PS: Consider   lapw7   for plotting abs/real/imag part of a wave fuction.
Remember: such phases are not unique, but can contain anarbitrary exp(i
phi) factor due to diagonalization.