This thread is locked.Only browsing is available.
Top Page > Browsing
LDA+U method
Date: 2007/04/05 02:52
Name: Allessandro Pennotti

Dear users,

I have looked over different examples shipped with OpenMX especially tests for LDA+U method. For example, in FeO+U.dat file U correction applied for 1d, because s2p1d1 basis set was used. But, in other examples, basis set for Fe is s2p2d2f1 (FeO_NC.dat). If I choose this basis set, in which d-orbital I have to apply U value? In other case, why f orbital was used in that example? In my opinion, s2p2d1 is enough for Fe.

Thank you very much for your help!
Best regards,
Page: [1]

Re: LDA+U method ( No.1 )
Date: 2007/04/05 09:01
Name: T.Ozaki


For the first d-orbital in the "d2" case, the U should be applied,
since the first d-orbital is the ground state corresponding to 3d orbital,
and the second d-orbital is the first exicited state with one node like 4d state.

As you pointed out, I agree that s2p2d1 is enough for Fe in most cases.
But when the orbital magnetic moment is calculated, a rather rich
basis functions including f-orbital are required for the convergent result.
You can easily check that the s2p2d1 basis gives a rather smaller orbital moment,
while the spin magnetic moment does not change so much (it would be better to calculate
the spin moment using the Voronoi analysis).



Page: [1]