[WARNING] DIPOLES database not correct or not present

Various technical topics such as parallelism and efficiency, netCDF problems, the Yambo code structure itself, are posted here.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano, Conor Hogan, Nicola Spallanzani

Post Reply
Bruno
Posts: 72
Joined: Tue Dec 08, 2020 11:16 am

[WARNING] DIPOLES database not correct or not present

Post by Bruno » Wed Dec 23, 2020 5:44 pm

Hello!

I'm doing some convergence test on a system and I got this error related to dipoles database. The yambo is compiled on top of a Quantum-Espresso 6.7. I was able to perform SCF and NSCF calculations and extract the SAVE folder with p2y with no problems. I initialized and generated the yambo input but I can't get through this error. Can somebody shed some light please?

In attachment you'll find the inputs for scf, nscf and yambo. And also the r_setup and the r- file generated by yambo accusing the mentioned error. It's worth to mention that I was able to run this system up to BndsRnXp = 700, but on another machine with the exact same inputs. Now I'm using another machine which lends me this error, even if the p2y seems work fine. Could this be a compilation problem on yambo? Or is something else?
yambo.tar.gz
Best regards,
You do not have the required permissions to view the files attached to this post.
MSc. Bruno Cucco
PhD Candidate
CNRS Institut des Sciences Chimiques de Rennes, France
Université de Rennes 1
https://iscr.univ-rennes1.fr

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: [WARNING] DIPOLES database not correct or not present

Post by Daniele Varsano » Wed Dec 23, 2020 6:03 pm

Dear Bruno,
note that the warning you report in the title (DIPOLES database not correct or not present) is harmless as it just warns that there is not a compatible dipole database present, so yambo calculate it.

It seems that yambo writes the database correctly but then it comes out with an I/O error, not easy to spot as I can't see anything weird in your input and report.

In order to try to understand what is going on:
*Can you check if the ndb.dipoles and ndb.dipoles_fragments* are all present in your 850b_1Ry directory?
*Can you also check if at the end of the logs file there is some more info that can help to spot the problem?
*Do the calculation runs smoothly if you include few bands (just a small test if it is a matter of a large number of bands you are using or it is independent on that)?

Maybe others, more expert than me on netcdf/hdf libraries can tell you more or suggest some test to do.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

User avatar
claudio
Posts: 448
Joined: Tue Mar 31, 2009 11:33 pm
Contact:

Re: [WARNING] DIPOLES database not correct or not present

Post by claudio » Wed Dec 23, 2020 11:02 pm

Dear Bruno

1) try to run in serial and see if you get the same error.

2) Second advice, generate the input with the -V par flag
and set parallelization of dipoles only in c and v not in k

3) go on github
https://github.com/yambo-code/yambo
and download the new version 5.0 that is online, even if we did not release it officially,
we corrected different bugs in this version

best
Claudio
Claudio Attaccalite
[CNRS/ Aix-Marseille Université/ CINaM laborarory / TSN department
Campus de Luminy – Case 913
13288 MARSEILLE Cedex 09
web site: http://www.attaccalite.com

Bruno
Posts: 72
Joined: Tue Dec 08, 2020 11:16 am

Re: [WARNING] DIPOLES database not correct or not present

Post by Bruno » Thu Dec 24, 2020 10:15 pm

Dear Daniele and Claudio,

Thank you very much for your help, I took both of your suggestion into account. It seems that setting the parallelization with -V par and setting c,v on dipoles actually solved the problem! I was able to run the calculation with BndsRnXp = 1 | 850 and NGsBlkXd up to 5 Ry with no problems so far. It also worked fine in serial so I suppose it was just a matter of setting the parallelization flags correctly as suggested (I had to work on the other ones too due to memory problems).

Again, thank you very much and Merry Christmas!

Regards,
MSc. Bruno Cucco
PhD Candidate
CNRS Institut des Sciences Chimiques de Rennes, France
Université de Rennes 1
https://iscr.univ-rennes1.fr

Post Reply