Open Shell Systems

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
vitoversace
Posts: 58
Joined: Thu Mar 13, 2014 3:43 pm

Re: Open Shell Systems

Post by vitoversace » Sun Apr 06, 2014 4:43 am

Dear Daniele,
Thanks a lot for your kind replies.

I have done the calculation for benzene and i get for instance for the 11th excitonic state :

Code: Select all

# Electron-Hole pairs that contributes to Excitonic State 11 more than 5%
#
#             K-point [iku]                      Weight
#   0.000000   0.000000   0.000000    0.127363
#   0.500      -.306E-16  -.306E-16      0.286    
#   0.00      0.500      -.306E-16          0.289    
#   0.500      0.500      -.612E-16        1.00    
#   0.000000   0.000000   0.500000    0.109865
#   0.500      -.306E-16  0.500            0.204    
#   0.000000   0.500000   0.500000    0.207410
#   0.500000   0.500000   0.500000    0.652437
#  
#  Band_V     Band_C     K  ibz     Symm.      Weight     Energy
#  
  15.00000   21.00000    4.00000    1.00000    0.24443    6.63483
  15.00000   21.00000    8.00000    1.00000    0.16787    6.79957
  15.00000   25.00000    3.00000    1.00000    0.06640    7.15014
  15.00000   25.00000    2.00000    1.00000    0.06576    7.15347
  14.00000   21.00000    4.00000    1.00000    0.06174    6.63822
  15.00000   23.00000    7.00000    1.00000    0.05596    7.24592
  15.00000   23.00000    6.00000    1.00000    0.05504    7.24994
So this one

Code: Select all

 15.00000   21.00000    4.00000    1.00000    0.24443    6.63483
at 4th k-point has the largest contribution, what is now the irreducible representation of the 11th excitonic state ?
How to deduce it now from symmetries of KS-states ???

To your information, benzene has 15 occupied states : with following irreducible representations from state 1 to 30 :

Code: Select all

        1A1g      1E1u      1E1u      2A1g      1E2g      1E2g      1B2u      1A2u      1B1u      2E1u      2E1u    
        2E2g      2E2g      1E1g      1E1g      1E2u      1E2u      2B1u      3A1g      1B2g      3E2g      3E2g
        3E1u      3E1u      4E2g      4E2g      4E1u      4E1u      3B1u      1A2g
Bests
Vito
Vito Versace
PhD student
University of York
vitoversace@yahoo.com

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

Re: Open Shell Systems

Post by Daniele Varsano » Sun Apr 06, 2014 7:42 am

Dear Vito,
1) if you want to calculate the excitation spectra of a molecule, you should use just the gamma point, unlike you are interested in a molecular crystal. Using just one k-point the picture in terms of weight should be simpler.
2) Symmetries of excited states: this is not a question related to the code. I just gave an hint on how to deduce them (when possible). Try to find a way for that in the literature or discussing with your supervisor or your group.

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/

vitoversace
Posts: 58
Joined: Thu Mar 13, 2014 3:43 pm

Re: Open Shell Systems

Post by vitoversace » Tue Apr 08, 2014 7:28 am

Dear Daniele,

Thanks a lot for your valuable hint, i now found a way to get the symmetry.

1) How can i have a much more dense grid for calculation of excitonic wave functions in ypp ??? (I tried the keyword CGrcSp but it did n't work!)
2) How can i get symmetric x-axis for excitonic wave function calculation in ypp, for instance from -10 to 10 ???

3)
Solvent effects: it is possible only in a very approximate way, considering the solvent as point charges, calculating the ground state in QM/MM framework using the cpmd code. Next yambo is interfaced with cpmd. If you plan to do that, you need to know few things that are not documented.
what should i already know and when would be approximately the next yambo release ?

Thank you very much and bests
Vito
Vito Versace
PhD student
University of York
vitoversace@yahoo.com

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

Re: Open Shell Systems

Post by Daniele Varsano » Wed Apr 09, 2014 4:30 am

Dear Vito,

1) Try to raise the FFTGvecs number in the input file.
2) The excitonic wvf is calculated in the unit cell or repeated unit cells. Selecting a gnuplot format, and the x-direction by
Format=g
Direction=1
you should have it. If you need to repeat cells, use an odd number. The hole will be placed in the central supercell. I think that yambo put the origin in the hole position, but I'm not sure. Anyway you can center it by hand and have your plot.
3) Yambo release: really do not know. I hope soon we can release a new release with MPI/OpenMP features.

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/

vitoversace
Posts: 58
Joined: Thu Mar 13, 2014 3:43 pm

Re: Open Shell Systems

Post by vitoversace » Wed Apr 09, 2014 4:07 pm

Dear Daniele,

1) When running the bse calculation, at the end i get the following warning :

Code: Select all

 <15m-01s> [07] BSE solver(s)
 <15m-01s> [WARNING] Missing non-local pseudopotential contribution
 <15m-01s> [07.01] Diagonalization solver
 <15m-01s> BSK diagonalize |                    | [000%] --(E) --(X)
 <15m-12s> P01: BSK diagonalize |####################| [100%] 10s(E) 10s(X)
And it affects my results. So how can i get this missing non-local pseudopotential contribution into my calculations?


2) Before running the bse calculation, i did a gw calculation on real axis, but when calculating Dipole (T) part, yambo tells me i should rename s.kb_pp file in SAVE otherwise the calculation will be slow because one has to
calculate a commutator.
Since i also wanted to have this commutator calculation in my calculations i DID NOT rename the s.kb_pp file and kept continuing but unfortunately yambo crashes at 50% of Dipole(T) calculation with the following error :

Code: Select all

Vito@Vito:~/Desktop/yambo/naphtalin$ time /usr/bin/mpirun -np 7 yambo -F Inputs/03_real_axis -J Real_Axis
   At line 142 of file io_bulk.f90 (unit = 42, file = './SAVE//s.kb_pp')
   Fortran runtime error: End of file
--------------------------------------------------------------------------
mpirun has exited due to process rank 6 with PID 12028 on
node vito exiting without calling "finalize". This may
have caused other processes in the application to be
terminated by signals sent by mpirun (as reported here).
--------------------------------------------------------------------------
I do not know how important this commutator calculation is really and i do not know why the non-local pseudopotential contribution is missing in BSE,
i did exactly what i also did for some other molecules in a simulation box and i had absolutely no problem. Strange !!!

Bests
Vito
Vito Versace
PhD student
University of York
vitoversace@yahoo.com

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: Open Shell Systems

Post by Davide Sangalli » Thu Apr 10, 2014 8:02 am

Dear Vito,
I tryed to look for that in the older post but there are too many.

Could you please recall me on which system you are working and attach to the reply your pwscf / abinit input file and pseudo-potential.

Best regards,
Davide
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/

vitoversace
Posts: 58
Joined: Thu Mar 13, 2014 3:43 pm

Re: Open Shell Systems

Post by vitoversace » Thu Apr 10, 2014 12:14 pm

Dear Davide,

Attached is my abinit inputs.
1) Every time i have a mesh size greater than 100 100 100 yambo complains in gw calculation on real axis and also at later bse calculation with those following errors.
I did exactly the same thing for the same test molecule but this time with a mesh size of 72 72 72 and every thing worked perfectly with very reasonable results without any warnings.
So effectively i can not investigate bigger molecules in a cubic simulation box larger than 30 Bohr.

my system is Linux 64 bit Ubuntu 10.04 LTS.

2) When i use the coupling keyword in BSE run for a test molecule in a box i get absolutely wrong results for the excited state energies but without it the results are correct.
I do not understand why this is the case ??? I always thought that coupling terms in BSE matrix are important for molecules.

Bests
Vito
You do not have the required permissions to view the files attached to this post.
Vito Versace
PhD student
University of York
vitoversace@yahoo.com

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

Re: Open Shell Systems

Post by Daniele Varsano » Fri Apr 11, 2014 6:34 am

Dear Vito,

1) Real-axis, please check in your standard output the memory you are allocating in your runs and how much memory you have in your machine. Real-axis calculations are very heavy (for this reason the plasmon-pole approximation has been introduced). The allocated memory depend linearly on the number of frequency you are considering. This is controlled by the ETStpsXd variable.
You can also play with the FFTGvecs number of vector to reduce your memory allocation, you need first to check in which part of the code it stacks.

2) Coupling is important for molecules, this does not means that it gives better results with respect Tamm-Dancoff approximation, it depends on the system. In many case the Tamm-Dancoff perform better. Usually in the case your KS ground state are not of very good quality (because of the functional) and the TD avoids to couple them. This for instances happens in polyenic conjugated molecules.

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/

Post Reply