GW calculation in 3.2.5-rev26 vs 3.2.5-rev34

[soh]

Hi,

I recently installed yambo 3.2.5-rev34 and had been previously using 3.2.5-rev26. I wanted to test the yambo -g n -p p calculation for a two atom Si cell. However the GW shift to the eigenvalues was wildly different for the same input files. Upon inspection the only difference in the report files was this:

rev 26
< Green`s function energies :Perdew & Zunger (xc)
< wavefunctions :Perdew & Zunger (xc)
< PPA diel. fun. energies :Perdew & Zunger (xc)
< wavefunctions :Perdew & Zunger (xc)
---
rev 34
> Green`s function energies :+Vosko, Wilk & Nusair (parametrization of the RPA energy)(C)
> wavefunctions :+Vosko, Wilk & Nusair (parametrization of the RPA energy)(C)
> PPA diel. fun. energies :+Vosko, Wilk & Nusair (parametrization of the RPA energy)(C)
> wavefunctions :+Vosko, Wilk & Nusair (parametrization of the RPA energy)(C)

I've attached yambo.in (which was generated with yambo -g n -p p), r_HF_and_locXC_em1d_ppa_gw0 and o.qp for both calculations. Is there some additional flag needed with rev 34 in yambo.in?

Many thanks,
Siobhan

Siobhan O'Halloran
PhD Student
Materials Theory Group
Tyndall National Institute
Cork
Ireland

[Daniele Varsano]

Dear Siobhan,
this happens because you are using a new revision with old database. The new revision does not read the previous databases correctly
and produce not reliable results. The Vosko, Wilk & Nusai you read in the report is a nonsense caused by the wrong reading of the database.
This is something we will fix soon. The problem is solved by producing all the database with the same revision of the code.
Cheers,

Daniele