Running a evGW0 calculation

[AzulineSky]

Hi there,

one step in the "Self-consistent GW on eigenvalues only" tutorial is not quite clear to me.
When running the input file g1w0_input.in, do you give the same -J option as before?
If you have to use a new directory like -J g1w0, can you reuse the dipoles for a self-consistent evGW0 calculation and save computation time?

Thanks for the help!

Best regards,
Max

[Daniele Varsano]

Dear MaX,

you can store the different ndb.QP for each iteration changing the -J directory at each step (G0W0, G1W0, G2W0 etc...).
The in the input file you indicate the QP you want to read obtained in the previous run: GfnQPdb= "E < ./G0W0/ndb.QP"
Dipoles, screening (in case you do not want or update it) databases etc., are anyway read as soon as they are in the SAVE directory.

Best,
Daniele

[AzulineSky]

I'm still a little confused. I did a G0W0 calculation for Si with reasonable parameters using

mpirun -np 64 yambo -F g0w0.in -J g0w0

If I now copy g0w0.in and add GfnQPdb= "E < ./g0w0/ndb.QP" to the input and rename it to g1w0.in and start Yambo with

mpirun -np 64 yambo -F g1w0.in -J g1w0

The calculations both take the same amount of time and the screening is recalculated in the G1W0 step.

What else am I doing wrong?

[Daniele Varsano]

Dear MaX,

you need to copy the databases you do not want recalculate (e.g. dipoles, screening etc.) in the SAVE directory, as Yambo will look for calculated quantities in the SAVE (always) plus the dir indicated with -J.

Best,
Daniele