ABINIT Discussion Forums

in the input file *.in, when computing the screening , can k points be selected arbitrarily?
the related parameters in the input file are listed as followed:

# Dataset3: Calculation of the screening (epsilon^-1 matrix)
optdriver3 3 # Screening calculation
getkss3 -1 # Obtain KSS file from previous dataset
nband3 100 # Bands to be used in the screening calculation
ecutwfn3 4.0 # Planewaves to be used to represent the wavefunctions
ecuteps3 6.0 # Dimension of the screening matrix
ppmfrq3 16.7 eV # Imaginary frequency where to calculate the screening
kptopt3 -3
ndivk3 10 12 17
kptbounds3 0.5 0.0 0.0 # This grid contains the Gamma point
0.0 0.0 0.0
0.0 0.5 0.5
1.0 1.0 1.0
istwfk3 40*1
awtr3 0 # Note : the default awtr 1 is better


error information is given in the *.log as follow:
......
m_bz_mesh.F90:1470:WARNING
k1 = 0.0000 0.0000 0.0000 is symmetrical of k2 = 1.0000 1.0000 1.0000 through sym = 23 itim = 2
m_bz_mesh.F90:1470:WARNING
k1 = 0.0000 0.0000 0.0000 is symmetrical of k2 = 1.0000 1.0000 1.0000 through sym = 24 itim = 2

m_bz_mesh.F90:1479:ERROR
input kibz is not irred.


please give your suggestion for this problem!
thank you in advance.

kptopt -3, kptbounds and ndivk are used to define the k-path for the band structure calculation
within the Kohn-Sham formalism in which the NSCF Hamiltonian can be
easily constructed at an arbitrary k-point starting from the density.
These variables are meaningless for GW in which the set of k-points for the QP corrections
is restricted to the k-mesh used to generate the KSS file.

You have to use the default kptopt=1, and the set of variables ngkpt, nshiftk and shiftk to define the BZ sampling
to be used for calculating the screened interaction and the self-energy. Remember that these values *must* be equal
to the ones used to generate the KSS file.

Matteo

gmatteo wrote:kptopt -3, kptbounds and ndivk are used to define the k-path for the band structure calculation
within the Kohn-Sham formalism in which the NSCF Hamiltonian can be
easily constructed at an arbitrary k-point starting from the density.
These variables are meaningless for GW in which the set of k-points for the QP corrections
is restricted to the k-mesh used to generate the KSS file.

You have to use the default kptopt=1, and the set of variables ngkpt, nshiftk and shiftk to define the BZ sampling
to be used for calculating the screened interaction and the self-energy. Remember that these values *must* be equal
to the ones used to generate the KSS file.

Matteo

thank you very much!
I think, what you said above is right. And from the help file and some related documents, some similar opinions are demonstrated. Indeed, in the scrrening calculation, probablely the regular k grid including some interesting k points needs to be given. the assignment of k points as provided above can't be performed in GW calculation in ab init.
so, express my thanks to you again for your provision.

is there any way to determine k-points in screening calculation

for example

we want to GW correct on 40 k-points and this points were divided four section

1 A point
2 Gama point
3 M point
4 L point

this point is in brillioun zone of ZnO wurtzide

and i want to obtain true band structure on this point

thank you