[Q-e-developers] About "molteplicities" in the resp_mat.f90 script

[huangxu1706 at sina.com]

Dear all, 
I'm confused about the part of "molteplicities" in the resp_mat.f90 script 
which is used to calculate U value of metal oxides. This part is after the 
construction of the first matrix = chi0m1 - chim1, and is important for the 
U-extrapolation from primitive cell to supercell. But I'm not sure the basic 
idea of this part, because it is beyond the theoretical introduction of the 
linear response (M. Cococcioni and S. de Gironcoli, Phys. Rev. B 71, 035105 
(2005)). 

For this part, I checked the output Umat.n.n.n.out files and found that the 
variables in the loops, i00, d0, mol, kp, kp0 and spinp, must have something 
to do with the "number of shells, shell #n: pc, sc, spinp dist chi_pc and 
chi_sc" in the output. But what is the physical meaning of those variables? 
How are they correlated with the chi0 and chi in primitive cell, and those 
in super cell? Why is the "molteplicities" necessary?

I'm working on a project about adapting the U-calculation which was designed 
for 3-D bulk system ny QE to 2-D surface system. I've made a lot of changes 
in the resp_mat.f90 script. I got reasonable U values in the first matrix = 
chi0m1 - chim1, but when it reached the molteplicities part, the calculated 
U values are strange. I believe that this part is also for 3-D system that 
some variables or loops also need to be modified for 2-D system. I can't treat 
it as a "black box" and trust its output. So I have to understand how this part 
works.

Thank you,
Xu Huang
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20140422/38144fce/attachment.html>