Large negative HF energies and QP correction with SOC

You can find here problems arising when using old releases of Yambo (< 5.0). Issues as parallelization strategy, performance issues and other technical aspects.

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

Locked
jlev1992
Posts: 5
Joined: Tue May 25, 2021 4:48 pm

Large negative HF energies and QP correction with SOC

Post by jlev1992 » Tue May 25, 2021 5:14 pm

Hi All,

In a recent calculation of G0W0 qp corrections using yambo v 4.5.1, I get large negative qp corrections when including spin-orbit coupling in QE (v 6.4.0). I've attached the report and qp output files. It looks like the band energies from QE are read in correctly, but the calculation of the HF energies seems off (large negative values). Do you have any advice for how to correct this? Thank you, and best regards!

-Josh
You do not have the required permissions to view the files attached to this post.
Joshua A Leveillee
Post-Doctoral Research Fellow
Oden Institute for Computational Sciences and Engineering
The University of Texas at Austin

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

Re: Large negative HF energies and QP correction with SOC

Post by Daniele Varsano » Wed May 26, 2021 9:06 am

Dear Josh,
please sign your post with your full name and affiliation, this is a rule of the forum and you can do once for all by filling the signature in you user profile.
Actually I can't see anything wrong in your input file.
May ask you to update the source to the latest release and repeat the calculation?

Code: Select all

git clone git@github.com:yambo-code/yambo.git yambo
in case we need to debug, we will do it in the latest release.

if the problem persists, please report here, adding all the needed file to reproduce your error (QE input file and pseudopotentials).

A side comment: you are using a psp with non-linear core correction. Even if this is allowed and yambo can take them into account when calculating the <Vxc> terms, they are discouraged as then the nlcc contribution is not taken into account in the HF terms, exactly for the same reason they are discouraged in QE when using hybrid functionals. Note anyway that this is completely unrelated with the problem you are experiencing.

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/

jlev1992
Posts: 5
Joined: Tue May 25, 2021 4:48 pm

Re: Large negative HF energies and QP correction with SOC

Post by jlev1992 » Wed May 26, 2021 9:11 pm

Dear Daniele,

My apologies for the lack of formal signature, I have updated this.

I now use the latest version of yambo from the master branch found at https://github.com/yambo-code/yambo. Additionally, I have used the latest formal release of QE found at https://github.com/QEF/q-e/releases/tag/qe-6.7.0.

Modules I use for compilation are intel/19.1.1 and impi/19.0.9

Configuration is done simply by ./configure FC=ifort F77=ifort

The machine I'm running on has Intel Xeon Platinum 8280 processors, in case hardware is a concern.

Yambo runs successfully but still produces large negative and nonsense QP correction with including SOC. There was a post years ago about perhaps a similar issue (viewtopic.php?t=825), though it's hard to tell if there was a conclusion.

I've attached IO below but the Ga psp file was too large to include, perhaps I can email this directly? Additionally I had to add the .txt extension to files to attach them.

Thank you for looking into this, it is greatly appreciated. Let me know what other information I can provide to help the investigation along. Cheers!

-Josh
You do not have the required permissions to view the files attached to this post.
Last edited by jlev1992 on Wed May 26, 2021 9:17 pm, edited 1 time in total.
Joshua A Leveillee
Post-Doctoral Research Fellow
Oden Institute for Computational Sciences and Engineering
The University of Texas at Austin

jlev1992
Posts: 5
Joined: Tue May 25, 2021 4:48 pm

Re: Large negative HF energies and QP correction with SOC

Post by jlev1992 » Wed May 26, 2021 9:13 pm

Looks like some previous files did not attached, including some more here, thanks!

-Josh
You do not have the required permissions to view the files attached to this post.
Joshua A Leveillee
Post-Doctoral Research Fellow
Oden Institute for Computational Sciences and Engineering
The University of Texas at Austin

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

Re: Large negative HF energies and QP correction with SOC

Post by Daniele Varsano » Thu May 27, 2021 8:54 am

Dear Joshua,

honestly it is not easy to spot the problem here, and most probably we should try to reproduce the issue.
Before doing that can you do these two tests:

1) can you see if the results do depends on the used parallelisation: can you repeat the calculation, maybe using less CPUs, and using a parallelisation strategy as:

Code: Select all

 | SE_CPU= "1 1 #cpu"                 # [PARALLEL] CPUs for each role
 | SE_ROLEs= "q qb b" 


2) compile the code using the option (--enable-int-linalg ) in the configure (after a make clean) and rerun it (after removìing the ndb* database).

For both the tests, a calculation of the HF part only is enough.

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/

jlev1992
Posts: 5
Joined: Tue May 25, 2021 4:48 pm

Re: Large negative HF energies and QP correction with SOC

Post by jlev1992 » Thu May 27, 2021 6:45 pm

Dear Daniele,

It looks like option 1 solved the problem! When I parallelize solely over bands as:

| SE_CPU= "1 1 #cpu" # [PARALLEL] CPUs for each role
| SE_ROLEs= "q qb b"

for the self energy, I get a reasonable HF energies and GW qp shifts. Thank you for the suggestions, I am relieved that this is working. Do you have an idea of why the parallelization choice effects the results when SOC is included?

Cheers!

-Josh
Joshua A Leveillee
Post-Doctoral Research Fellow
Oden Institute for Computational Sciences and Engineering
The University of Texas at Austin

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

Re: Large negative HF energies and QP correction with SOC

Post by Daniele Varsano » Fri May 28, 2021 7:53 am

Dear Josh,
at the moment I do not have a clue, and this is something we are going to investigate.
Thanks for reporting.

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
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Large negative HF energies and QP correction with SOC

Post by Davide Sangalli » Fri May 28, 2021 1:45 pm

Dear Josh,
I'm trying to reproduce the issue.

I need the UPF of Ga.

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/

jlev1992
Posts: 5
Joined: Tue May 25, 2021 4:48 pm

Re: Large negative HF energies and QP correction with SOC

Post by jlev1992 » Fri May 28, 2021 10:44 pm

Dear Davide and Daniele,

The Ga UPF file is too large to attach, is there an email address you prefer I send it to? You can email me off-forum at joshua.leveillee@gmail.com if you prefer to not share it here. Cheers!

-Josh
Joshua A Leveillee
Post-Doctoral Research Fellow
Oden Institute for Computational Sciences and Engineering
The University of Texas at Austin

Locked