The ABINIT Users Mailing List ( CLOSED )

[Jorge Iniguez <jiniguez@nist.gov>]

Dear abinitioners,

A couple of weeks ago Gilles Z. mentioned the possibility of doing
calculations with magnetization constrained to lie along a given spatial
direction (see emails about Berry curvature & Chern number sent by H. Min
as well as GZ's response).

I think it is worth to implement such option in abinit. At the very least,
it would allow us to check whether the spinor calculations are working
properly.

The following is a summary of results for MnF2 (2 formula units per cell),
for different values of spinat (the first two atoms are the Mn atoms).
These are calculations with nspinor=2 and nspden=4 (see attached input
file for details; note that all calculations are started from scratch,
i.e. no wave functions are read). The calculations are not converged in k
points and cutoff, but the results should still be meaningful. The
energies are converged to 1.e-10 hartree.

spinat1  0.0  0.0  5.0   0.0  0.0  5.0   12*0.0
 ETOT 32  -125.26182581685    -1.418E-11 6.879E-04 1.226E-07 1.478E-06 
2.799E-05
 3D magnetization:     -4.271522E-02  -5.674532E-02   9.999220E+00

spinat2  0.0  0.2  4.8   0.0  0.2  4.8   12*0.0
 ETOT 32  -125.26182581081    -2.075E-11 8.996E-04 1.926E-07 2.089E-06 
2.956E-05
 3D magnetization:     -4.033234E-02   3.366962E-01   9.993805E+00

spinat3  0.0  0.5  4.5   0.0  0.5  4.5   12*0.0
 ETOT 32  -125.26182580903    -1.246E-11 1.005E-03 1.652E-07 1.431E-06 
3.249E-05
 3D magnetization:      4.283887E-02   1.085839E+00   9.940317E+00

spinat4  0.0  0.21 4.8   0.0  0.21 4.8   12*0.0
 ETOT 32  -125.26182581099    -2.173E-11 8.465E-04 1.931E-07 2.073E-06 
2.965E-05
 3D magnetization:     -4.116870E-02   3.586754E-01   9.993034E+00

spinat5  0.0 -0.2  4.8   0.0  0.2  4.8   12*0.0
 ETOT 33  -125.26182657667     3.368E-12 5.127E-04 5.037E-06 1.708E-06 
3.512E-05
 3D magnetization:     -3.912451E-02  -1.140459E-01   9.987464E+00

As you see, I give very similar "seeds" with spinat... but get five
different answers. The AF modification in spinat5 (which is the way to
reduce the system energy, since the system ground state is AF) is
completely missed by the first four runs. (Note that this problem will be
present even if we do not have spin orbit. An spinor calculation should be
able to find the ground state (FE or AF) of the system, right?) Regarding
the first four cases, one could be tempted to think that the energy
differences involved are so small that the run gets converged before
actually finding the minimum (in spite of the very small tolerance used).
However, case 5 involves relatively large energy differences and the
program fails to go all the way up to an AF ordering. This seems to imply
something is not working as it should with the spinor calculation? (If
somebody thinks the above results can be easily explained and there is
nothing to worry about, please let me know :)

Being able to set the magnetization direction would be helpful to
understand what is happening here. (Maybe the above result is simply
indicating that the energy landscape of the system is very complicated? We
could answer that question if we were able to fix the direction of M.) It
would also allow to calculate magnetic anisotropies, etc.

I would very much like to work in the implementation of this.

What do you think?

Jorge


------------------------------------------------------------------------------
   Jorge Iniguez             Bldg 235, Room E19
                             NIST Center for Neutron Research
                             100 Bureau Drive, Mail Stop 8562
                             Gaithersburg, MD 20899-8562
                             Phone: 1-301-975-8367 Fax: 1-301-921-9847
                             E-mail address: jiniguez@nist.gov
                             http://www.ncnr.nist.gov/staff/jorge
------------------------------------------------------------------------------# Structure

acell 9.2026119831  9.2026119831  6.3036905801
rprim 1.0 0.0 0.0  0.0 1.0 0.0  0.0 0.0 1.0

natom 6
ntype 2
znucl 25 9
type  2*1 4*2
      xred    0.0000000000E+00  0.0000000000E+00  0.0000000000E+00
              5.0000000000E-01  5.0000000000E-01  5.0000000000E-01
              3.0584104371E-01  3.0584104371E-01  0.0000000000E+00
             -3.0584104371E-01 -3.0584104371E-01  0.0000000000E+00
              1.9415895629E-01  8.0584104371E-01  5.0000000000E-01
              8.0584104371E-01  1.9415895629E-01  5.0000000000E-01

# xc, cutoff and k points

ixc 1   # LDA
ecut 1000 eV
kptopt 1
kptrlatt 2 0 0   0 2 0    0 0 4
nshiftk 1
shiftk 3*0.5


# Electrons

nspden 4   ; nsym 1
nspinor 2  ; pspso 2 2
ndtset 5
spinat1  0.0  0.0  5.0   0.0  0.0  5.0   12*0.0
spinat2  0.0  0.2  4.8   0.0  0.2  4.8   12*0.0
spinat3  0.0  0.5  4.5   0.0  0.5  4.5   12*0.0
spinat4  0.0  0.21 4.8   0.0  0.21 4.8   12*0.0
spinat5  0.0 -0.2  4.8   0.0  0.2  4.8   12*0.0

#irdwfk 1

nband 50
occopt 3
tsmear 0.002

# Calculation options

nstep 200     
toldfe 1.0d-10