[Pw_forum] multiple-manifold DFT+U on a single atom

[Deyu Lu]

Dear all:
         I'm interested in doing DFT+U calculation on a single atom with multiple
manifolds, and I'm wondering whether the code is available in some developmental stage.
To be more specific, I'm looking at small Ce clusters and want to understand how linear 

response U of Ce changes according to the local chemical environment, e.g., valence state
and spin state, etc. If I use only Ce 4f, I found chi0 and chi matrices don't obey sum rule,
and in some cases, an eigenmode of chi0 and chi can be clearly identified corresponding to 

a uniform potential perturbation with a small negative eigenvalue. This means there is an exchange

of the induced charge density between the 4f manifold and the background. 

By looking at PDOS, a non-negligible  4f-5d and 4f-6s hybridization are found right at LUMO, 

however, the hybridization with nearby O 2p seems to be weaker. In order to eliminate this error 

during the chi0 and chi inversion and
obtain more accurate estimate on U, it makes sense to me to add Ce 5d and 6s. This modification
is not quite obvious in set_hubbard_l.f90 and tabd.f90, since these subroutines assume one channel
per atom type.

         Best regards,
         Deyu Lu
************************************************
Deyu Lu
Assistant Physicist, Theory & Computation Group
the Center for Functional Nanomaterials
Rm 1002, Building 735, Brookhaven National Lab
Upton, NY, 11973
webpage: http://www.bnl.gov/cfn/people/Deyu_Lu.asp ************************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20120130/09af526a/attachment.html>