The ABINIT Users Mailing List ( CLOSED )

["Matteo Giantomassi" <Matteo.Giantomassi@uclouvain.be>]

> Hi.
> I have found the problem. That is the pseudopotential file. I don't
> understand why because I can run GS calculation normally with these
> pseudopotential files. I have changed pseudopotential file and now can run
> GW.
>
> Best Regarded
> Quan Phung Manh
>

 Dear Quan,

 What kind of pseudopotentials are you using and which version of abinit?
 Old versions of abinit do not generate the KSS file if
 the pseudopotentials contain spin-orbit coupling or if at least one pseudo
 has more than one projector per angular channel.
 In this case the code writes out a comment in the log file and skips the
calculation
 of the KSS file. You might look for the comments reported in the log by
the routine "outkss"
 to make sure that your pseudopotentials are compatible with the your
version of Abinit.

 By the way: I had a look at your input file and I have a comment on the
value of nbandkss.
 You are using kssform=1 (default value) and nbankss=-1 to generate the
KSS file.
 This means that the band structure is generated through a FULL
diagonalization of the Hamiltonian.
 Have a look at these links for more details:

 http://www.abinit.org/Infos_v5.5/input_variables/varfil.html#kssform
 http://www.abinit.org/Infos_v5.5/input_variables/vargw.html#nbandkss

 The size of the Hamiltonian matrix is defined by ecut (40 Ha in your case)
 thus the full diagonalization should result in thousands of bands to be
calculated, quite CPU demanding!
 I don't think you need such a large number of states for your GW
calculation; as
 a rule of thumb the number of empty states should be one order of
magnitude larger than
 the number of occupied bands. To speed up the calculation you can specify
the number of states
 that have to be explicitly calculated and stored in the KSS file. For
example, you can
 start with nbandkss 300-500 to perform the first convergence tests.

 Hope it helps,
 Best Regards
 Matteo Giantomassi