NaN in PPA calculation

Concerns issues with computing quasiparticle corrections to the DFT eigenvalues - i.e., the self-energy within the GW approximation (-g n), or considering the Hartree-Fock exchange only (-x)

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano

User avatar
Davide Sangalli
Posts: 640
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: NaN in PPA calculation

Post by Davide Sangalli » Wed Nov 18, 2015 4:50 pm

Ciao,
the bose temperature is used when the bose distribution of the plasmon is computed.
It is used as a smearing parameter, more than a really physical temperature.
The default (-1) is simply an internal standard of yambo.
In this case yambo sets BoseTem="electronic temperature"
Try to play with it using the fast run (i.e. the input file with small parameters) if you think it may affect the result

Anyway, I did the same run with the GPL stable but I did not get any NaN. ( I also used BoseTemp=0.00 )
The differences maybe due to:
a) different configuration flags in compilation.
b) a slightly different revision

Thus, can you enter the folder which contains the yambo source and:
a) Take the file "report" inside the config folder and attach it here
b) run svn info and tell me which is the number you read nearby "Revision: XX"

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

mferri
Posts: 17
Joined: Thu Nov 12, 2015 10:18 am

Re: NaN in PPA calculation

Post by mferri » Thu Nov 19, 2015 11:52 am

Hi Davide,
Here you are the Revision Number and the report file.
Revision: 90
Node Kind: directory
Schedule: normal
Last Changed Author: sangalli
Last Changed Rev: 80
Last Changed Date: 2015-05-15 16:27:38 +0200 (Fri, 15 May 2015)
Cheers,
Matteo
You do not have the required permissions to view the files attached to this post.
Matteo Ferri
PhD student in Condensed Matter at SISSA

Previously:
Master at Università degli Studi di Milano &
CNR, Istituto per la Microelettronica e Microsistemi - Sezione di Agrate Brianza

User avatar
Davide Sangalli
Posts: 640
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: NaN in PPA calculation

Post by Davide Sangalli » Thu Nov 19, 2015 1:02 pm

Dea Matteo,
we spot a problem which maybe the source of your NaN.

Try this. Open the file src/common/FREQUENCIES_coarse_grid.F, remove the line

Code: Select all

if (df>tresh) i2=i2+1
recompile yambo and do again the same run.

Do you still get the NaN ?

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

mferri
Posts: 17
Joined: Thu Nov 12, 2015 10:18 am

Re: NaN in PPA calculation

Post by mferri » Fri Nov 20, 2015 10:04 am

Dear Davide,
thank you for the advice. Because of some technical issues I think I cannot do the calculation until Monday.
I will let you know the result as soon as possible.

Best,
Matteo
Matteo Ferri
PhD student in Condensed Matter at SISSA

Previously:
Master at Università degli Studi di Milano &
CNR, Istituto per la Microelettronica e Microsistemi - Sezione di Agrate Brianza

mferri
Posts: 17
Joined: Thu Nov 12, 2015 10:18 am

Re: NaN in PPA calculation

Post by mferri » Mon Nov 23, 2015 10:22 am

Hi Davide,

I modified the file FREQUENCIES_coarse_grid.F as you said, and now there are no more NaN. Using the same input I got the QP corrections.
# K-point Band Eo E-Eo Sc(Eo)
#
1.00000 1.00000 -10.98441 0.08640 7.28282
1.000000 2.000000 -9.210250 -0.270751 5.923223
1.000000 3.000000 -7.527469 -0.529275 4.913851
1.000000 4.000000 -6.048070 -1.010020 4.023360
1.000000 5.000000 -3.446251 -0.709965 3.388270
1.000000 6.000000 -2.640253 -0.772790 3.193753
1.000000 7.000000 -2.352882 -0.874099 3.072914
1.00000 8.00000 -1.47106 -0.02367 2.70648
1.000000 9.000000 -0.323451 0.260259 2.015086
1.00000 10.00000 0.00000 0.27504 2.17540
1.00000 11.00000 1.23334 1.17054 -1.01305
1.00000 12.00000 1.96749 1.22927 -1.91303
1.00000 13.00000 3.85810 1.27017 -2.26053
1.00000 14.00000 3.97166 1.01944 -2.08620
1.00000 15.00000 4.10548 1.50865 -2.08123
1.00000 16.00000 4.28350 1.70597 -2.14227
1.00000 17.00000 4.56151 0.35905 -1.59270
1.00000 18.00000 4.74564 1.51291 -2.22349
1.00000 19.00000 5.89140 0.53087 -1.09958
1.00000 20.00000 5.90880 0.77628 -1.87733
Thank you very much for your help.
Best,

Matteo
Matteo Ferri
PhD student in Condensed Matter at SISSA

Previously:
Master at Università degli Studi di Milano &
CNR, Istituto per la Microelettronica e Microsistemi - Sezione di Agrate Brianza

User avatar
Davide Sangalli
Posts: 640
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: NaN in PPA calculation

Post by Davide Sangalli » Mon Nov 23, 2015 11:13 am

Nice.

Thank you for spotting the problem. We have released a commit in the repository.
Next time you do "svn up" to upgrade the source the change will be automatically included.
You do not need to modify it again or to remember where the change was in the source.

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

Post Reply