The ABINIT Users Mailing List ( CLOSED )

[Valerio Olevano <valerio.olevano@polytechnique.fr>]

Dear all,
I'd like to say some words on the subject, as it's
asked a lot and to not return on it in the future:
in principle one can calculate GW corrections only on
the k-points in the set of your selected grid.
But a grid suitable for a good ground state calculation
might not contain the meaningful k-points for
a band structure calculation (e.g. Gamma, X K etc, as well as
the points along meaningful directions, Delta etc.).
So, once you have done your ground state calculation and get
a well converged GS density, a good advise is to select the more
complete
grid containing these meaningful k-points, and do the GW calculation
on these. 
Another good advise could be to use shifted grids (shiftk)
which allows also to reuse as screening the same EM1 SCR files 
as obtained in the non shifted screening calculation. 
And shift many times so to cover all the points you need.
But somehow you encounter some problems,
the shiftk is not working really in this sense, and usually you have
to turn off symmetries.
It's very intricate to get a GW bandstructure, but I think
that until also a calculation of the LDA/GGA bandstructure 
will not be really straightforward and simple (something like put
an option bndstrc 1 in the input and getting at the end
an agr file containing the picture for the bandstructure
with all the important points and directions for that BZ shape
corresponding to that space group), well I think it's a nonsense to
ask for something similar for the GW one.
Cheer up!
Valerio
> to handicaps prevent you from performing full bandstructure gw
> calculation:
> 1, of course the time. Even single point may take you several hours.
> 2, not all points you define in the input file can be calculated.
> There is a restriction on the point, due to symmetry.
> Regards
>  Rongqin Wu
> 
> -----Original Message-----
> From: quantumfancy@hotmail.com [mailto:quantumfancy@hotmail.com] 
> Sent: 2004年4月22日 23:33
> To: forum@abinit.org
> Subject: [abinit-forum] About GW
> 
> Dear all,
> 
> I am conducting electronic structure calculation on BP.In my
> works,Hamann scheme norm-conserving pseudopotentials are used,together
> with the Perdew-Burke-Ernzerhof generalized gradient
> approximation(GGA) is adopted.I obtain an indirect gap of 1.28 eV.As
> expected,it differs significantly from the experimental value about
> 2.0 eV.
> 
> It seems that the GW approximation may be a better choice.I have gone
> through the sixth lesson of tutorial and Test_v3/t30.in and t31.in,but
> in these examples,only the band gaps and GW corrections of some
> specific k-points are given.My question is whether I can employ the GW
> approximation implemented by Abinit to calculate the entire band
> structure?Any help would be greatly appreciated. 
> 
> Many thanks,
> 
> Li Huang
> 
>