About magnetic system

Message

look · #1 Post by **look** » Wed Jun 01, 2005 1:02 pm

When doing relaxations of magnetic atom on nonmagnetic atoms surface, I set ISPIN=2 and big value of MAGMOM which was 8 for magnetic atom and 0 for nonmagnetic atoms in the INCAR file. But it still converged to a nonmagnetic state which had mag=0.02 in the OSZICAR file. Does someone know the reason?

#2 Post by **admin** » Mon Jun 13, 2005 9:23 am

sometimes the convergence to the magnetic solution is better if you
-- start from the non-magnetic groundstate,
-- keep CHGCAR and WAVECAR and
-- continue with
ICHARG = 1
ISPIN =2
set in INCAR.
Have you tried this approach as well?

lahaye · #3 Post by **lahaye** » Wed Oct 25, 2006 7:35 am

With VASP 4.6.28, I have done following:

1) relaxation of the system without spin-polarization.

2) calculate accurately total energy (ISMEAR=-5 etc.) without
spin-polarization; and I kept CHGCAR and WAVECAR for next run.

3) ran again, with only change in INCAR: ICHARG=1 and ISPIN=2.

The output of VASP then says:

Code: Select all

found WAVECAR, reading the header
nup: number of bands has changed, file:  28 present:  39

Should I worry about this changed number of bands?
Should I run everything again with NBANDS = 39
enforced manually in INCAR?

Thanks,
Rob.

#4 Post by **admin** » Wed Oct 25, 2006 12:20 pm

if any of the calculations parameters do not match, reading the WAVECAR file will be skipped. If you want to read it though, you have to set NBANDS=39 in the non-spinpolarized run explicitely.
However, you can as well continue from CHGCAR only if you start spin-polarization.

bandy · #5 Post by **bandy** » Mon Nov 13, 2006 4:56 pm

I am confused about how to reach magnetic state from self-consistently converged nonmagnetic solution. As the user "look" got nonmagnetic solution even though he used ISPIN=2, in that case does he need to start the calculation with ISPIN=2 ICHARG=1 or he first needs to get ISPIN=1 nonmagnetic state and then start from this nonmagnetic state by ISPIN=2 ICHARG=1? In this case as Rob said NBANDS needs to be set by hand. If to start from nonmagnetic state reached by ISPIN=2 calculation, how a calculation starting from this solution will be different as CHGCAR already converged with ISPIN=2 condition.

Thanks for any clarification

<span class='smallblacktext'>[ Edited Mon Nov 13 2006, 06:02PM ]</span>