How to calculate full dielectric tensor within GWA

[roginovicci]

I know it is possible to get the macroscopic dielectric constant at the calculation of the screening step. That is supposed to be an average value, right? But how to extract each component (xx,yy,zz,xy,xz,yz) of dielectric tensor within GW approximation? Is it possible?

[roginovicci]

just to make it clear. The components of dielectric tensor could be separately calculated by setting gw_qlwl variable ( for instance 0.001 0 0 for \varepsilon_xx) and inspecting spectra in _EM1_LF _EM1_NLF files