Strange behaviour with spin, but only in supercell

Deals with issues related to computation of optical spectra, solving the Bethe-Salpeter equation.

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

Post Reply
FalkoSchmidt
Posts: 2
Joined: Tue Jul 24, 2018 2:19 pm

Strange behaviour with spin, but only in supercell

Post by FalkoSchmidt » Tue Aug 21, 2018 1:14 pm

Dear Yambo community,

since I am currently working on a rather large system (polaron at a point defect in a lithium niobate supercell with 80 atoms), I started by calculating the linear response for two test systems, both without point defects and polarons (stoichiometric): The primitive cell with 10 atoms and without spin (system "1" in my notation), and the 2x2x2 repetition with 80 atoms with spin ("29"). Additionally, I used two waypoints: A 10 atom cell with spin ("3") and an 80 atom cell without spin ("18").

Please have a look at the attached figure (plot.pdf): "Switching on" the spin in the small cell ("1"->"3") does not change the result noticeably, as expected. Neither does changing to the supercell ("1"->"18"). Technically, there are changes, but those had to be expected, since I used less k-points. The strange behaviour occurs, when we try to calculate the supercell including spin: The linear response is way too large (factor 4) and its form does not fit too well, either.

Im am not certain how to explain this, since the energies in chapter 02.05 of the file r-IPA_optics_chi are very similar for systems "18" and "29".

Attached you will find all files I used for my calculations. I would be grateful for any suggestions.

Best,
Falko
You do not have the required permissions to view the files attached to this post.
Falko Schmidt
Department of Physics
University of Paderborn, Germany

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: Strange behaviour with spin, but only in supercell

Post by Davide Sangalli » Tue Aug 21, 2018 4:21 pm

Dear Falko,
your result is indeed strange.

Can I ask you to try the same calculation (just cases 18 and 29) neglecting the non local part of the pseudo in the calculation of the dipoles, to check if the problem comes from there.
The calculation will become significantly faster. You just need to rename the file SAVE/ns.kb_pp_pwscf

Code: Select all

mv SAVE/ns.kb_pp_pwscf SAVE/ns.kb_pp_pwscf_hide
before running the same input files with a different jobstring name.

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/

andrea.ferretti
Posts: 206
Joined: Fri Jan 31, 2014 11:13 am

Re: Strange behaviour with spin, but only in supercell

Post by andrea.ferretti » Wed Aug 22, 2018 10:09 am

Dear Falko,

I've examined your files and it seems that the failure (in terms of numerical results) of your supercell spin polarized run (#29) is due to a p2y bug that we have recently spotted and fixed.
In particular, here is a description of the bug:

* the bug affects yambo v4.2.0-v4.2.2 interfacing with QE > 6.2.x, if
- QE is not compiled against HDF5 and the QEXSD format is used (both options are the default)
- p2y fragments wfc files over bands (default is only over kpts, unless large systems are treated, as is your case here)
or p2y -b xxx is used.
In practice, if only files of the form SAVE/ns.wf_fragment_X_1 are present, p2y works fine, while if any
SAVE/ns.wf_fragment_X_Y (Y>1) is there, the bug shows up.

* the bug has been fixed in yambo v4.2.3

Hope it helps
Andrea
Andrea Ferretti, PhD
CNR-NANO-S3 and MaX Centre
via Campi 213/A, 41125, Modena, Italy
Tel: +39 059 2055322; Skype: andrea_ferretti
URL: http://www.nano.cnr.it

FalkoSchmidt
Posts: 2
Joined: Tue Jul 24, 2018 2:19 pm

Re: Strange behaviour with spin, but only in supercell

Post by FalkoSchmidt » Wed Aug 22, 2018 3:19 pm

Dear Davide and Andrea,

upgrading to the newest version of Yambo did indeed solve all of my problems. Thanks a lot for your help!

Best,
Falko
Falko Schmidt
Department of Physics
University of Paderborn, Germany

Post Reply