DOS in noncollinear calculation?

Message

Veronika · #1 Post by **Veronika** » Mon Oct 24, 2005 7:52 am

Hm, which LORBIT did you set? Maybe try LORBIT = 11 or 12 ...

Veronika · #2 Post by **Veronika** » Fri Nov 04, 2005 7:01 am

If you run with vasp 4.6 or higher you should use the integer numbers for LORBIT as described in the manual (http://cms.mpi.univie.ac.at/vasp/vasp/node113.html) ... The exact interplay between LORBIT and LNONCOLLINEAR I do not know but I suggest to set LORBIT if you want to get more information on the DOS ...

windiks · #3 Post by **windiks** » Fri Nov 04, 2005 10:43 am

Dear all,

the terms "spin up" and "spin down" do not exist in the concept of noncolinear magnetism or spin-orbit calculations. You obtain a magnetisation density only as defined in the paper of Hobbs, Kresse and Hafner PRB 62(17), 1156 (2000). The paper explaines how the concept of noncolinear magnetism within the local spin density theory is implemented in VASP.
As far as I know, the DOS is calculated from that magnetization density and, hence, you do not get a DOS for spin-up and spin-down states. Furthermore, the tag ISPIN has nothing to do with the tags LORBIT and LNONCOLLINEAR!
But you can calculated noncolinear magnetic moments without taking into account spin orbit coupling (i.e. LNONCOLLINEAR=.TRUE. and LORBIT=.FALSE. is possible).

Again, do not mix the terms "spin-polarized calculations" with "calculations of noncolinear magnetism". Both calculations are based on completely different concepts!

Regards,
RenÃ©

#4 Post by **admin** » Wed Nov 16, 2005 1:13 pm

for non-collinear magnetism, the second block of data in DOSCAR
contains
E, and 4 columns for each, s,p,d, giving:
rho, m_x, m_y, m_z
with m....magnetisation

lahaye · #5 Post by **lahaye** » Tue Dec 26, 2006 12:20 am

I did a NON-Collinear / LSORBIT=.TRUE. / ISPIN=2 calculation,
and the second block in DOSCAR has for each atom NEDOS
rows with 37 columns. The first column is clearly the DOS energy,
but what are the other 36 columns?

Thanks,
Rob.

#6 Post by **admin** » Wed Dec 27, 2006 10:57 am

my last posting answers your question.
Energy , 9x4 entries for the decomposed local DOS of each atom.

As already pointed out by Rene (2 postings above), it makes absolutely NO SENSE to set ISPIN=2 (up and down) for non-clollinear runs, therfore this tag is ignored when it s read from INCAR.

wellsdm · #7 Post by **wellsdm** » Mon Feb 11, 2008 9:28 pm

I have been doing f-electron LS-coupling runs can you explain the make-up of the DOS in that case? i have what seems to be 37 columns (E+rho,mx,my,mz for each s,p,d) then a second line of another 28 before i get another line starting with an understandable energy value. Can you please explain this. if the second line is the f-electron components, why is the first number formatted more like energy (0.000 instead of 0.0000E+00 like the others)

jermysmart · #8 Post by **jermysmart** » Wed May 25, 2011 11:27 am

whether I can get a spin-up-DOS and spin-down-DOS invovling SOC by projecting DOS to up and down

chelman · #9 Post by **chelman** » Thu Oct 27, 2011 5:44 pm

Dear head admin: Following this thread, I would like to know if the file vasprun.xml contain the same information that DOSCAR. if it's true, then p4v can read it? how?
Thanks!

#10 Post by **admin** » Fri Oct 28, 2011 1:04 pm

Yes. p4v/control/Electronic/LocalDOS+bands control

yywang · #11 Post by **yywang** » Mon Apr 01, 2013 6:40 am

[quote="jermysmart"]whether I can get a spin-up-DOS and spin-down-DOS invovling SOC by projecting DOS to up and down[/quote]

I also want to know the answer, and how to project DOS to up and down. Has your problem been solved?