The ABINIT Users Mailing List ( CLOSED )

[Johann Bouchet <bouchet@pcpm.ucl.ac.be>]

Dear Ab-init users,

I'm actually using GW for cells with a quite large numbers of atoms (12 
or more) and as you know the main problem is the time of the 
calculations. For GW we have to check the convergence of 6 parameters (3 
for chi, 3 for sigma). Even if some of them don't have too much impact 
on the time, you have to do at least 12 calculations and it's even 
closer to 20. In the tutorial the convergence is done only for the gamma 
point. I did the convergence studies of a two atoms system for 1 and 19 
k points, it seems that I obtain the same results, I mean for the value 
of the parameter, for the gap I have a big difference. Of course it's a 
lot faster with only the gamma point! So I wanted to know if someone as 
an experience about that? Can we do the convergence studies only for the 
gamma point and after that just run one BIG calculation with the right 
number of k points?
For me it's working, but I just did one test.

thanks,

Johann