SPIN-Orbit calculation dumps only one spin channel in EIGENVAL, DOSCAR files

Problems running VASP: crashes, internal errors, "wrong" results.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
salva
Newbie
Newbie
Posts: 11
Joined: Thu Oct 12, 2006 6:57 pm
License Nr.: 592
Contact:

SPIN-Orbit calculation dumps only one spin channel in EIGENVAL, DOSCAR files

#1 Post by salva » Mon Oct 16, 2006 7:45 pm

Hello,
I am new with the code so I might be seeing this as a bug that is not really present. I'm using VASP 4.6.28, compiled with pgi fortran and running an a xeon cluster with cent-Os (free fedora)

I have made a calculation on bulk gold both with ISPIN=1 and 2. ISPIN=2 gives me results for two spin channels, as expected.

Then I turn on the SPIN-ORBIT coupling. And the output dumps only one eigenchannel. My INCAR file follows:
===========================================================
Bulk gold test calculation
ISPIN=2
SYSTEM = AuBulk
NWRITE = 2
PREC=Accurate
IALGO=48
GGA=91
LSORBIT = .TRUE.
SAXIS = 0 0 1 ! direction of the magnetic field
NBANDS = 50
NUPDOWN = 1
===========================================================
Standard output is just normal:
[salva@tbird AuBulk]$ ./vasp
running on 1 nodes
distr: one band on 1 nodes, 1 groups
vasp.4.6.28 25Jul05 complex
POSCAR found : 1 types and 1 ions
LDA part: xc-table for Ceperly-Alder, standard interpolation
generate k-points for: 12 12 12
|| W W A A R R N N II N N GGGG !!! || || Full k-point grid generated || Inversion symmetry is not applied || | -----------------------------------------------------------------------------

POSCAR, INCAR and KPOINTS ok, starting setup
WARNING: wrap around errors must be expected
FFT: planning ... 1
reading WAVECAR
entering main loop
N E dE d eps ncg rms rms(c)
.
.
.
RMM: 18 -0.350209045025E+01 0.16626E-04 -0.10425E-06 32337 0.238E-03 1 F= -.35020905E+01 E0= -.35020912E+01 d E =0.224021E-05 mag= 0.0003 0.0003 0.0003
writing wavefunctions
FORTRAN STOP
===========================================================

These are the first entries of the EIGENVAL file:
1 1 1 1
0.1825866E+02 0.2955706E-09 0.2955706E-09 0.2955706E-09 0.5000000E-15
1.0000000000000000E-004
CAR
AuBulk
11 364 50

0.3742002E-15 0.3742002E-15 0.3742002E-15 0.5787037E-03
1 -4.2850
2 -4.2849
3 -0.1777
4 -0.1776
5 -0.1774
6 -0.1774
===========================================================
...showing only one spin component.

What do I need to do to override the "Startparameters" as dumped by OUTCAR?

Startparameter for this run:
NWRITE = 2 write-flag & timer
PREC = accura medium, high low
ISTART = 0 job : 0-new 1-cont 2-samecut
ICHARG = 2 charge: 1-file 2-atom 10-const

ISPIN = 1 spin polarized calculation? <=======This particular line!

LNONCOLLINEAR = T non collinear calculations
LSORBIT = T spin-orbit coupling

==========================================================

I haven't found a way around the problem. Your assistance will be greatly apprecciated.

Thank you in advance,
S.
Last edited by salva on Mon Oct 16, 2006 7:45 pm, edited 1 time in total.

admin
Administrator
Administrator
Posts: 2921
Joined: Tue Aug 03, 2004 8:18 am
License Nr.: 458

SPIN-Orbit calculation dumps only one spin channel in EIGENVAL, DOSCAR files

#2 Post by admin » Wed Oct 18, 2006 12:25 pm

If you switch on SOC, the notation 'spin up' and 'spin down' does not make sense, the "spin-components" are split in 4-spinors, giving the total charge and the magnetisation components with respect to the principal axis. Please have a look at the LSORBIT chapter of the online manual (http://cms.mpi.univie.ac.at/CMSPage/main) for further details
Last edited by admin on Wed Oct 18, 2006 12:25 pm, edited 1 time in total.

salva
Newbie
Newbie
Posts: 11
Joined: Thu Oct 12, 2006 6:57 pm
License Nr.: 592
Contact:

SPIN-Orbit calculation dumps only one spin channel in EIGENVAL, DOSCAR files

#3 Post by salva » Wed Oct 18, 2006 7:29 pm

Thank you for the prompt reply. While waiting I did read section 6.59 on non-collinear spin and spin-orbit coupling. This places my post in the physics question section...
I understood the point that now VASP is dealing with spinors. In such a case I get an eigenvalue per spinor, and not two if I wanted to think of spin-up and down components. Is this correct?

I would like to say what I want to do. Say I want to obtain the band structure of gold with spin-orbit corrections. I would like to get the *band splittings* that accompany the onset of spin-orbit interactions. Can VASP perform such a calculation?

--Having this application in mind, would it be correct to just build those band structures from two runs, one with the magnetic moment parallel to the spin quantisation axis, while the second run has the magnetic moment antiparallel to this axis? this is the only thing I can think of, but I am not sure this is really going to be the answer to my question. I would appreciate if someone more experienced could comment on that.--

Besides the section in the manual, where could I find more about it (a research article would be very helpful)?

Thank you for your consideration,
Salvador.
Last edited by salva on Wed Oct 18, 2006 7:29 pm, edited 1 time in total.

Post Reply