ndb.kindx

[xjxiao]

Dear all,
I have saved the ndb.pp* obtained from the G0W0 calculation into 120b_15Ry_240Gb_bse (hoping that subsequent calculations can read the static dielectric function from it). However, when running the calculation with `yambo -o b -k sex -F bse1.in -J 120b_15Ry_240Gb_bse`, I noticed that it modifies the SAVE/ndb.kindx. Is this an issue? Will this affect other calculations?
Thanks!
Xiao

[Daniele Varsano]

Dear Xiangjun,

this is not an issue, the database has been regenerated with the needed indexes and the grid is recognized as uniform, so everything should be ok.

Kind regards,

Daniele

[xjxiao]

Dear Daniele,
If I change parameters or perform other types of calculations, can I continue to use this modified SAVE folder? Thanks!
Xiao

[Daniele Varsano]

Dear Xiao,

yes, YAMBO reads all the databases store in the SAVE directory, plus optional directory indicated with the -J option.
In the report file you can find all the database that have been read indicated with "RD" and the new ones written on disk "WR".

In general, if databases are found, and they are not compatible with the input file parameters they are recalculated.
In some cases, if the incompatibility is not judged as severe, they are not recalculated but a warning "WRN" is emitted.

For convergence test you can use the -J option saving databases in different directory and keep in the SAVE only the ones that will not change.

Best,

Daniele

[xjxiao]

Dear Daniele，
Sorry, I’m still a bit confused.

For convergence test you can use the -J option saving databases in different directory and keep in the SAVE only the ones that will not change.

During the kernel calculation, the SAVE folder gets modified, and I can’t keep it unchanged. Can the modified SAVE folder still be used for G0W0 calculations or BSE calculations with different parameters?
Thanks！
Xiao

[Daniele Varsano]

Dear Xiao,

sorry, but I do not get your problem.

In SAVE you have the wavefunctions ns.*, the database generated by the setup (ndb.gops and ndb.kindx) and any other database Yambo generates.
Yambo always read the SAVE. If database are compatible with the input they are used, if they are not, they are recalculated and overwritten.

If you want to keep databases with different parameters you can use the -J option.
e.g.

yambo -J 200bands5Ry
yambo -J 400bands5Ry
etc..

In this way, the new database will be stored in different directories with the given name.

They are also read in following calculations if the run indicates the -J directory.

Best,

Daniele

[xjxiao]

Dear Daniele,
Thank you for your patient replies, but I keep encountering issues when generating BSE-related input files and running BSE calculations.

First Attempt:
Yesterday, I placed the ndb.pp* files obtained from the G0W0 calculation in the folder 120b_10Ry_240Gb_bse.
Then, I ran the command:

Code: Select all

yambo -o b -k sex -y d -V qp -r -F bse_rim.in -J 120b_10Ry_240Gb_bse/  

The calculation failed without any error messages. What could be the problem?

Second Attempt:
Today, I reused the previously modified SAVE folder (even though it was altered during the last calculation). I placed the same ndb.pp* files in a new folder 120b_10Ry_240Gb_bse2 and tried the same command:

Code: Select all

yambo -o b -k sex -y d -V qp -r -F bse_rim2.in -J 120b_10Ry_240Gb_bse2/  

This time, the generated input file lacked some parameters. Why did this happen?

[Daniele Varsano]

Dear Xiao,

you have a problem with the k point index in the BSE:

Code: Select all

 X grid is uniform                                : no 

I suggest you to start a calculation from scratch, adding the flag NoDiagSC in the setup and see if the ndb.kindx is generated correctly.
If it fails, check your k point grid: avoid user defined k point grids, shifted grids and if possible ibrav=0 in your QE nscf input file.

Best,

Daniele

[xjxiao]

Dear Daniele,
Thanks!
In my setup, I already included NoDiagSC, and the GW calculation completed successfully. The k-grid for GW appears to be uniform. Even though the BSE calculation failed, are my previous G0W0 computation trustworthy?

I did use ibrav=0 in my NSCF calculation. Now, I plan to restart.

Xiao

[Daniele Varsano]

Dear Xiao,

the GW calculation should be ok.
If you manage to calculate your ground state using a defined ibrav this should help Yambo in recogninzing a uniform k point grid.

Best,

Daniele