merged database not read by yambo

[rreho]

Dear developers,
I am trying to calculate the BSE spectra reading the GW corrections from a merged database with Yambo 5.0.
First I calculate the GW corrections for multiple couples of bands and then I merged the database with ypp.
After that I use bse.in reading from the merged database.
However, Yambo is not able to read the database where the corrections are stored. I reproduced the error in hBN and I attach here the input/report/log files.
Thank you for your help and your time.

Best regards.
Riccardo Reho

Utrecht University, Netherlands

[Daniele Varsano]

Dear Riccardo,

the correct syntax to read the database in the use input file is:

Code: Select all

KfnQPdb= "E < ../gw_merged/gw_merged_morebands/ndb.QP"                  # [EXTQP BSK BSS] Database action

there is a missing "<" in your input.

Best,

Daniele

[rreho]

Dear Daniele,
Thank you for your reply. I am sorry but in order to give you a general input file I deleted that character.
This is the original input file that I used for the calculation with the real path. As you can see I used the '<'.
Looking forward to your reply.

Best,
Riccardo

[Daniele Varsano]

Dear Riccardo,


many thanks for reporting?

indeed there is an I/O problem in reading the database. We need to inspect to understand what went wrong.
1) Please verify that the needed file exists in the indicated directory:
/gpfs/scratch/pr1eme00/pr1eme09/hBN/new/gw_merged/gw_merged_morebands/ndb.QP
2) If so, as I believe, can you post the ndb.QP file?
3) If we cannot spot any particular problem we need to reproduce you error.


Best,

Daniele

[rreho]

Dear Daniele,
This is the ndb.QP database obtained after I merged with ypp. As you can see it is named by default as 'ndb.QP_merged_1_gw_ppa'. I already tried to call it from the bse part both with his original name and after renaming it with ndb.QP. I thought it could have been some small mistake like that.

Thanks

Riccardo

[Daniele Varsano]

Dear Riccardo,
many thanks, we will inspect it and probably we will need to reproduce the problem.
In the meanwhile, as it seems the calculations are not heavy I suggest you to repeat a calculation including all the QP correction in a single run so that you can avoid to merge databases.

Best,
Daniele

[rreho]

Dear Daniele,
Thank you for your help. This was a test system. The actual system is a supercell of 42 atoms with a 9x9x1 k-mesh and spin-orbit.
I wanted to report here in the Forum the test system that we used in the school to help with finding the error.

Best,
Riccardo

[Daniele Varsano]

Dear Riccardo,
ok i see. We will try to reproduce the error and fix it.

Daniele

[Daniele Varsano]

Dear Riccardo,

unfortunately I was not able to reproduce your error and the database is read correctly, next the code stop as in your input file there an inconsistency between BSENGBlk (4Ha) and NGsBlkXs (4Ry). Anyway the merged database is read correctly, also when using the same database you post.

I used the last GPL release.

Note that the same problem you are experiencing i.e.:

Code: Select all

[WARNING]I/O problems in reading QP database 

I have also found when I misspelled the name of the file, so the file was not found.

So what I can suggest you is:
* Update to the newest release ( Version 5.0.3)
* be sure your are addressing the correct file, to be 100% sure you can set in the SAVE directory and setting KfnQPdb= "E < ./SAVE/ndb.QP_merged_1_gw_ppa" or in the same directory you are running omitting the SAVE. In any case the name of the file is not an issue.

Best,
Daniele

[Daniele Varsano]

Dear Riccardo,

it is possible that the length of the string is causing problems:

/gpfs/scratch/pr1eme00/pr1eme09/hBN/new/gw_merged/gw_merged_morebands/ndb.QP

you should try to provide a relative path instead of the absolute path e.g. if you are running in the directory,

/gpfs/scratch/pr1eme00/pr1eme09/hBN/new/
you can assign:
./gw_merged/gw_merged_morebands/ndb.QP

Best,
Daniele