The ABINIT Users Mailing List ( CLOSED )

[Fabien Bruneval <fabien.bruneval@polytechnique.fr>]

Dear pprodhom,

It is not correct to select some k-points for your KSS file. Your KSS 
file should contain all k-points on a regular grid (that can be reduced 
 by symmetry to the k-points in the irreducible Brillouin zone).
Indeed the code needs to know all the k-points and all the q-vectors on 
grids with the SAME spacing.
Of course, you can evaluate Sigma (optdriver 4) for the k-points you are 
interested in, but all the preparation runs have to deal with all the 
k-points.
Bonne chance.

Fabien



pprodhom@chartreuse.cea.fr wrote:
> Hi abiniters,
> I am doing calculation using the GW correction.
> I begin by set my k points in the BZ for the self-consistent calculation, 
> then I choose some K points on which I do a non self consistent calculation 
> to get the KS structure on this K points.
>
> Then during the epsilon^-1 calculation when abinit fourier transform the RPA 
> polarizability it should fix (to take into account the periodicity of the 
> density) the q point to ki-kj+G (G being a reciprocal lattice vector, ki and 
> kj are the Kpoint of the KSS) but what I see is that it just set the qpt to 
> kj-k1
>
> So at the end of the calculation when it calculates the self energy, it 
> checks if q is still equal to ki-kj+G, with kj the K points of KSS and ki the 
> point where we want to make the correction. But at that time if ki is 
> different from k1, it gives the error message bellow:
>
>  calculation status (     4 to be completed):
>   1
>   2
>   3
>  **error: q = k - kp + g0 not found.
>  kmkp  =    0.5000000000000000       -0.5000000000000000
>     0.000000000000000
>
> So I was wondering if I could correct the first equation when q is set equal 
> to kj-k1 and to make it equals to ki-kj so that at the end it won't give an 
> error message. I don't tink I am wrong if I do that, but if I am, could 
> anyone tell me what I miss?
>
> Thanks in advance
> Best regards