[Pw_forum] a question about LDA+U

[rjxiao]

Dear Matteo Cococcioni,

I did what you have told me.It does work!But directly inserting the old version of the ns_adj routine in the new code will lead an error.There are four lines which should be changed.
---------
f(m1,m2) = nsnew(na,mins,m1,m2)---->f(m1,m2) = nsnew(m1,m2,mins,na)
f(m1,m2) = nsnew(na,majs,m1,m2)---->f(m1,m2) = nsnew(m1,m2,mins,na)
nsnew(na,adjs,i,j) = dreal(temp)---->nsnew(i,j,adjs,na) = dreal(temp)
nsnew(na,adjs,j,i) = nsnew(na,adjs,i,j)---->nsnew(j,i,adjs,na) = nsnew(i,j,adjs,na)
--------------
Maybe the two versions use different order of parameters in the function nsnew.
By the way, I compare the ns_adj.f90 file in espresso2.1 with the earlier version(such as pwscf2.0.4 and pwscf2.0),I can not find any difference.And the espresso2.1 can not give an energy gap in calculation of Fe2O2.So,maybe this file should also be inserted to the espresso2.1 package.

I still believe that my coordination is not wrong.When it is shown in xcrysden,it can give the same picture as your coordination.And it gives the same result in the calculation as yours.I check these two kinds of coordinations carefully.I found that we put the cell with different orientation in the space and put the origin in different site of the cell.In your model, the Fe1(0,0,0) is the at the center of the cell,while in mine,the Fe1(0,0,0) is the atom on the peak of the hexahedron.

I am really thank you for your work.I think LDA+U is a very interesting and useful method.
 
Best regards,

Sincerely,
Ruijuan Xiao
Institute of Physics,
Chinese Academy of Sciences