[Pw_forum] EXX routine

[Helen]

I've been using the EXX routine to calculate electronic structure. The routine can calculate the orbital energies at the k-points used in the scf calculation. In an effort to calculate the electronic structure for a general k-point I have been adding the general k-point with an infinitesimal weight to the list of k-points used in the scf calculation along with its symmetry points e.g for nq=4,4,1 
K_POINTS
   18
   0.0000000  0.0000000  0.0000000   0.03125
  -0.2500000  0.2500000 -0.2500000   0.25000
   0.5000000 -0.5000000  0.5000000   0.12500
   0.0000000  0.5000000  0.0000000   0.18750
   0.7500000 -0.2500000  0.7500000   0.75000
   0.5000000  0.0000000  0.5000000   0.37500
   0.0000000 -1.0000000  0.0000000   0.09375
  -0.5000000 -1.0000000  0.0000000   0.18750
    0.0000000  0.0000000  0.001000   0.0000001
   0.2500000  0.2500000  0.25100   0.0000001
   0.5000000  0.5000000  0.501000   0.0000001
   0.7500000  0.7500000  0.751000   0.0000001
   0.0000000  0.0000000  0.501000   0.0000001
   0.2500000  0.2500000  0.751000   0.0000001
   0.5000000  0.5000000  1.001000   0.0000001
   0.0000000  0.0000000  1.001000   0.0000001
   0.25       0.25       1.251      0.0000001
   0.0        0.0        1.501      0.0000001

This method works fine for LDA routines. It also works OK with the functional we made (BNL, which is a type of hybrid functional of LDA and explicit exchange) when the weight given to the explicit exchange is small. However when I use only the explicit exchange routine or the EXX routine has a more significant weight than LDA, I get spurious results for the general k-point.
For example when q=4,4,1 for bulk Si
I get at k=0,0,0  orbital energies 3.3785eV and 11.6304eV
but at k=0,0,0.001 orbital energies 2.7821eV and 11.8583eV
when the energies at k=0,0,0.001 should be almost the same as those at k=0,0,0
This problem doesn't occur when q=1,1,1 and occurs to a lesser extent when q=2,2,1. The results for the 'normal' k-points are fine the problem is only with the general k-point.

Can anybody (especially Dr. de Gironcoli) explain this problem?
Thank you for your help,
Dr. Helen Eisenberg,
Fritz-Haber Center for Molecular Dynamics,
Hebrew University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20090607/36501b96/attachment.html>