LSDA+U can switch spin-unpolarized to spin-polarized,possible?

Message

yuxiao_zihan · #1 Post by **yuxiao_zihan** » Tue Apr 01, 2008 2:15 pm

It is my first time to try LSDA+U. with LDA (GGA), ISPIN=2, I get a spin-unpolarized band structure. But if I switch to LSDA+U, with U, J, ISPIN, I get a spin-polarized band structure. Is it possible?

Thanks.

#2 Post by **admin** » Thu Apr 03, 2008 1:43 pm

no, if you set ISPIN=2, you automatically switch on spin-polarization. If your system is spin-polarized in its ground state, the spin up and down states are not degenerate in energy (they are of course if your system is non-magnetic). Therefore it is sufficient just to set ISPIN=2.

yuxiao_zihan · #3 Post by **yuxiao_zihan** » Thu Apr 03, 2008 4:31 pm

Thanks for your reply.
Yes, I totally agree with you. Here I want to know what L(S)DA change. Ruthenium (Ru) is non-magnetic. It is true with LDA or GGA.

But with L(s)DA+U:
LDAU = .TRUE.
LDAUTYPE = 1
LDAUL = 2
LDAUU = 4.0
LDAUJ = 0.6
LMAXMIX = 4
LDAUPRINT = 2
ISPIN = 2

I get magnetic in OUTCAR
# of ion s p d tot
----------------------------------------
1 0.011 -0.006 1.523 1.529
2 -0.012 -0.016 -0.226 -0.254
3 -0.013 -0.018 -0.314 -0.345
4 0.010 -0.006 1.412 1.417
------------------------------------------------
tot -0.004 -0.045 2.395 2.346

What does it mean? Here the slab contains 4 layers 1x1 Ru. and K points is 36x36x1.

What happened with L(S)DA? What's my wrong?
Thanks
<span class='smallblacktext'>[ Edited Thu Apr 03 2008, 06:46PM ]</span>