DFT+U dE[U]/dP(i,j)

[Matthias Krack]

Dear Marco,

just check Eqs. 5 and 6 of Dudarev et al. PRB 57(3), 1505 (1998). Please, 
note the Kronecker delta in Eq 6.

Best regards,

Matthias


On Monday, 9 February 2015 19:02:24 UTC+1, Marco wrote:
>
> Hello,
>
> Sorry about this simple question but I was hoping someone could explain to 
> me how the potential is calculated in the DFT+U Mulliken method.
> In lines 1379-1383 of the dft_plus_u.f90 file, the potential dE[U]/dP(i,j) 
> is defined as:
>
> IF (isgf == jsgf) THEN
>   v_block(isgf,isgf) = u_minus_j*(0.5_dp - fspin*q_matrix(i,i))
> ELSE
>   v_block(isgf,jsgf) = -u_minus_j*fspin*q_matrix(j,i)
> END IF
>
> Why is the v_block(isgf,jsgf) not defined as:
> v_block(isgf,jsgf) = u_minus_j*(0.5_dp - fspin*q_matrix(j,i))
>
> Best regards,
> Marco
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150209/0d07dda2/attachment.htm>