[Wien] Using afminput for rutile MnF2

[Cormac McGuinness]

Dear list,

I thought it would be straightforward to setup an AFM calculation for
rutile MnF2, but I am clearly misunderstanding the manual and the
section regarding the afminput.  Clearly I can follow the example in
Dufek et al (PhysRevB 48, 12672, 1993) and do a spin-polarised
calculation for MnF2 using a cell of spacegroup Cmmm which is double
the size of the rutile until cell (spacegroup P42/mnm), and whose
a-axes are rotated 45 degrees from those of the tetragonal rutile unit
cell.  However, the local coordinate frames for all the atoms change
from the rutile unit cell to the Cmmm unit cell, as the global unit
cell basis it is rotated by 45 degrees. The spin up and down of the
now inequivalent magnetic ions (the Mn) are specified as input to
instgen_lapw

Instead of simply doing a spin-polarised calculation, I though I
should use afminput. Thus I have tried to follow the Wien2k userguide
and use afminput to set up a symmetry operation to map the spin up
states onto the spin down states, saving computer time and preserving
symmetry (as explained in sections 4.5.4 and 9.5). The expectation for
the symmetry transformation required is a 90 degree rotation about the
c-axis and a translation of 0.5, 0.5,0.5.

In doing so I have tried to set up the rutile unit cell for MnF2 as a
case.struct_supergroup in order to let afminput automatically select
the required symmetry operation and generate the Fourier components
for the transformation. However when running afminput  I keep getting
told that the "The super and subgroups are TRANSLATIONENGLEICH" and
get an identity matrix and a null primitive translation (0 0 0) as the
symmetry operation.
There is a specific warning in section 4.5.4 about cases where e.g. y
and z axes are interchanged at the behest of sgroup. Though this is
not the case in rutile, both the global coordinate frame is rotated,
and some of the local coordinate frames are also rotated from the
P42/mnm unit cell to the Cmmm unit cell. I have tried to heed these
warnings, but did not understand the exact sequence required (set
NSYM=0 in the case.struct but not the case.struct_supergroup?/  change
the local coordinate frames in all cases?)

Alternatively, in the absence of a case.struct_supergroup I have tried
to specify the rotation and translation directly, but to no avail.
afminput fails with an error "rrot not found", as no equivalent atom
is found in the Cmmm spacegroup for the transformation specified.

Finally, in frustration I tried taking the rutile call and converting
it into a P cell and splitting the Mn (and F) positions, and giving
the two Mn atoms u and d spins respectively through instgen_lapw.
Using the non-magnetic rutile cell as the case.struct_supergroup then
"x afminput" gave me the expected symmetry operation! However it also
gave me a zero length case.inclmcopy_st file


I have not attached anything to this request, as I had generated so
many struct files (not to mention .inst and .outputeafminput) files
that I would not want to burden the list with unnecessary attachments.

Should I just stick to the spin-polarised calculation?

Yours
Cormac

--
Dr. Cormac McGuinness, Lecturer, School of Physics., Trinity College,
Dublin, Ireland.
Phone: +353 1 8963547    Fax: +353 1 6711759    Office: SNIAM 3.17