Magnetism

starting a magnetic calculation

When you use the input-generator, this program tries to set up a magnetic calculation, whenever atoms like Fe, Co, or Ni appear in the input. Then jspins=2 is set in the inp-file and at the bottom a line like 

swsp=F  2.20  1.60  0.90

appears, indicating that the starting density should have magnetic moments of 2.2, 1.6 and 0.9 Bohr for the atom types 1, 2 and 3. If you want to start with different moments, e.g. an antiferromagnetic calculation, change this line like 

swsp=F  2.20 -2.20  0.00

then atoms 1 and 2 have antiparallel spins. Of course, the input generator does not know, that NiAl is nonmagnetic, Gd2Fe is antiferromagnetic, or a thin film of Rh is magnetic. This you have to specify yourself.

spin-polarizing a calculation

If you started from a non-magnetic calculation and want to spin-polarize your calculation,

  • make sure that you have a file "cdnc" in your working directory and set jspd=2 in the fl7para file (or simply remove it),

  • set jspins=2 and swsp=T in the inp-file, followed by the magnetic moments you want to induce in the atoms (see above).

  • running fleur.x results in STOP 'spin polarized density generated'

  • Now set again swsp=F in the inp-file and converge

If you want to flip the magnetization of some atom n in a spin-polarized calculation, set l_flip=T and set the n'th digit in the same line to -1. Run fleur.x and after the STOP set l_flip=F again.

Non-CollinearMagnetism

A short primer in how to do calculations using Non-Collinear Magnetism.

  1. First of all, converge a collinear ferromagnetic configuration.
  2. Then you should copy cdn1 to rhomat_inp.
  3. Next create a nocoinp-file (make sure, that it is in the right format!). If you want to use the constraint it is probably best to switch it on right from the start (l_mperp=T,l_constr=T). A good choice for mix_b is 0.5.
  4. You should now delete fl7para and the broyd-files.
  5. Then set l_noco=T in inp.
  6. If ctail=T in inp, set ctail=F.
  7. Then converge as usual noticing that rhomat_inp is now used instead of cdn1.

Be careful

when you set l_ss=T !

  1. This cannot be combined with constraints (l_constr=T) because it's not implemeted yet.
  2. When setting l_soc=T a perturbative scheme is used as the generalized Bloch theorem does not hold in the presence of spin-orbit coupling.

when you set l_mperp=T ! Then two things do not work correctly:

  1. the orbital moments in the out- and inf-file get wrong
  2. the density matrix (in file n_mmp_mat) is wrong

these things still have to be implemented to work together.

For an introduction into the technique, see the Psi-k Highlight FLAPW goes Non-collinear