The ABINIT Users Mailing List ( CLOSED )

[ธนูสิทธิ์ บุรินทร์ประโคน <thanusit@kku.ac.th>]

Dear Carlos Ortiz

I wonder if you have properly performed the convergence tests of the GW results with respected to nband, ecutwfn and ecutsigx used  for sigma calculation, and to nband, ecutwfn, ecuteps used for the screening calculation. In your input file, the values of the nbandkss and nband, ecutfwn, ecutsigx, and ecuteps seem to be too small for a good convergence of the GWs. This may be the cause of the strange results you have got.

Regarding the error "cchi0 : ERROR - kp=k-q-G0 not found", I used to encounter the similar problem. According to Matteo Giantomassi’s comments posted earlier in the forum, this error arises when the difference between k-q and k is not a reciprocal lattice vector. As a result, in screening calculation we seem to be limited to the k-mesh that is specified using the "standard" procedure, namely using ngkpt and nshiftk and shiftk, just like the one you have in your input file.

 Hope this gives some help

 

Kind regards,

Thanusit Burinprakhon

On June 12, 2008 6:00:03 AM ICT, "Carlos Ortiz" <caortiz76@gmail.com> wrote:
> hi,
> I am trying to get the GW values at G=[0 0 0], M=[1/2 0 0] and K=[1/3 
> 1/3 0] for graphite with abinit 5.4
>
> (see input/output files attached to this mail)
>
> with
> ngkpt 2*6 1
> nshiftk 1 # number of shifts hexagonal lattice
> shiftk 0.0 0.0 0.0
> i get 7 k-points
> 1 0.00000000E+00 0.00000000E+00 0.00000000E+00
> 2 1.66666667E-01 0.00000000E+00 0.00000000E+00
> 3 3.33333333E-01 0.00000000E+00 0.00000000E+00
> 4 5.00000000E-01 0.00000000E+00 0.00000000E+00
> 5 1.66666667E-01 1.66666667E-01 0.00000000E+00
> 6 3.33333333E-01 1.66666667E-01 0.00000000E+00
> 7 3.33333333E-01 3.33333333E-01 0.00000000E+00
> and use the plasmon pole approximation with omega
> 1 0.000000E+00 0.000000E+00
> 2 0.000000E+00 7.500000E+00
>
> my concern is that i get a crazy values (too big?) for
>
> #1 chi0(q= 1, omega= 1, G,G'), and corresponding epsilon(G,G') at 
> 1th and 2nd omega
> (also dielectric constant missing: dielectric constant = ******** 
> dielectric constant without local fields = ********)
> and
> #2 chi0(q= 7, omega= 1, G,G') and corresponding epsilon(G,G') at 
> the 1th omega
> (see graph.log file)
>
> why is this? does this mean i have unreliable GW results? (LDA gaps 
> change with -0.016, 0.325 and 0.023 at G, M and K respectively)
> if there is an error, is it coming from a divergency at G? if so, is 
> there a way to shift the k-point grid to contain K but not G and M?
>
> i have tried different shifts but my attempts have so far always ended 
> at
> cchi0 : ERROR - kp=k-q-G0 not found
> during the screening calculation
>
> regards
> carlos