Yambo Community Forum

Dear Daniele,

I ran GW calculations on LiF in yambo_3.4.1 and yambo_3.4.2, and obtained gaps with deviations of 0.2 eV (using exactly the same parameters in both yambo) !

Results of yambo_3.4.2 and yambo_4.1.2 are fully consistent.
However yambo_3.4.1 gives a gap which is larger by 0.2 eV relative to yambo_3.4.2 and yambo_4.1.2.

I further add that the results of yambo_3.3.0 and yambo_3.4.1 are fully consistent.

Furthermore, if you set parameters to converged values, deviation increases even more!

Since there are many publications based on yambo_3.4.1 and earlier versions of yambo, i would like to know what is going wrong here eventually.

Please find attached the corresponding data and outputs.

Best
Luca

Dear Luca,
thanks for posting this. We will investigate.
Can you post also the report file for the 3.4.1?
I noticed that the input files are not generated with the same release of the calculations, they are all done with 4.0.4,
if some variable named changed between releases it can cause problems.
Many thanks,

Daniele

Dear Daniele,

Sorry, i forgot to post it.
Attached the report file of yambo_3.4.1.

I paid attention that the variables are consistent within each release.


Best
Luca

Thanks,
we will have a deep look into that.
Most or all publications are based on plasmon-pole approximations, anyway, we will search carefully the reason for such a discrepancy and let you know.

Best,

Daniele

Dear Luca,
what happens if you set in both calculations "CGrdSpXd= 0.0000" ?

Best,
D.

Dear Davide,

I still obtain the 0.2 eV deviation in gap.
CGrdSpXd= 0.0000 changes nothing.

Best
Luca

Dear Luca,

thanks to Davide he went back in the log file of the yambo commits and he realized that indeed there was a bug in the previous releases that was fixed in 3.4.2.

Fortunately, the bug affected the real-axis implementation only and not the plasmon-pole approximation procedure, where most of the QP calculations are done.

Thanks again for pointing in out.

Best,

Daniele

Dear Daniele,

many thanks for clarification.

Now i need to redo all of my calculations with yambo_4.2.1.
Any way at least i know what is correct.

Best wishes
Luca