GW inconsistency ?

[martinspenke]

Dear Daniele,

I did the same GW calculation with yambo_3.4.1 and yambo_4.0.1_rev_88 but obtained different quasi particle energies !!!

[martinspenke]

and i think the results obtained from yambo_3.4.1 should be the correct ones.

By the way what happened to this variable

Code: Select all

% EnRngeXd
  0.00000 | 10.00000 | eV    # [Xd] Energy range
%

in yambo_4.0.1 ??? Even if i explicitly give the variable by hand in GW input file of yambo_4.0.1, i don't see any report of the energy variable in the output file !

Best wishes
Martin

[Daniele Varsano]

Dear Martin,
thanks for reporting,
can you send also the report file for the 3.4.1 calculation?

The EnRngeXd should be fixed to the min max energy given by Ec-Ev, but I need to be check it.
Best,
Daniele

[martinspenke]

Dear Daniele,

Sorry, i forgot to include the report file, too.
Attached the report file.

May i ask you the correct keyword for EnRngeXd in yambo_4.0.1 ?

Is it this one : ?

# | %QPerange # [GW] QP generalized Kpoint/Energy indices
# | 1| 36| 0.0|-1.0|
# | %

Unfortunately a keyword for energy range is not generated automatically in yambo_4.0.1.

Best wishes
Martin

[Daniele Varsano]

Dear Martin,
thank you, I will check it as soon as possible.
As I told you before EnRngeXd it is not needed anymore, if you check in 3.4 too, independently of the input, in the report you can see at the end of the report that it is automatically set to min(Ec-EV),max(Ec-eV). QPerange it is not related at all. Before I'll go through the report and eventually reproduce the error, let me ask you wether you are using OpenMPI libraries to compile yambo, and if this is the case, are you adding the flag --enable-openmpi ?

Best,
Daniele

[vitoversace]

Dear Daniele,

I have to say i checked that and obtained the same problem too.

Regards
Vito

[martinspenke]

Dear Daniele,

Yes, I added the flag --enable-openmpi at compilation time because i saw this issue in an another post by you.
I also used a sequential version of mkl.

Bests
Martin

[Davide Sangalli]

Dear Martin,
sorry for the late reply.

We find out that the differences were due to a bug in the parallelization of the gpl/devel (4.0)
We have presently fixed the bug in the *non* gpl version. As soon as the fix will be released also in the gpl we will let you know here.

D.

[martinspenke]

Dear Davide,

Thanks for your reply and many thanks for fixing the bug which unfortunately forced me stop using the 4.0 version of Yambo.

Any chance of releasing non equilibrium mbpt in near future ?
"Ultra-fast carriers relaxation in bulk silicon following photo-excitation with a short and polarized laser pulse"

Best wishes
Martin

[Davide Sangalli]

For the non-equilibrium mbpt part we will probably release that as well under GPL one day ... but not in the near future I believe.

Regards,
D.