Error in BSE calculation :BSE solver: diagonalization

Deals with issues related to computation of optical spectra in reciprocal space: RPA, TDDFT, local field effects.

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

Post Reply
Abeshipour
Posts: 11
Joined: Mon Mar 22, 2021 10:45 am

Error in BSE calculation :BSE solver: diagonalization

Post by Abeshipour » Sun Aug 01, 2021 10:31 am

Dear Yambo developers,

I am trying to calculate the dielectric function of NaBO2 (in bulk form) using BSE. I did my calculations according to this tutorial in the Yambo website: "Calculating optical spectra including excitonic effects: a step-by-step guide'

I am trying to converge using BS bands.
When the kernel is created and goes to next step, i. e., solving the BSE, I get an error as follows:
//
ambo-5.0.3/bin/yambo -F 03_3D_BSE_diago_solver.in -J 3D_30-65_BND
Invalid MIT-MAGIC-COOKIE-1 key

__ __ ____ ___ ___ ____ ___
| | |/ | | | \ / \
| | | o | _ _ | o ) |
| ~ | | \_/ | | O |
|___, | _ | | | O | |
| | | | | | | |
|____/|__|__|___|___|_____|\___/


<---> [01] MPI/OPENMP structure, Files & I/O Directories
<---> [02] CORE Variables Setup
<---> [02.01] Unit cells
<---> [02.02] Symmetries
<---> [02.03] Reciprocal space
<---> [02.04] K-grid lattice
<---> Grid dimensions : 5 5 5
<---> [02.05] Energies & Occupations
<---> [03] Transferred momenta grid and indexing
<---> [04] External/Internal QP corrections
<---> [QP_apply] Action to be applied: USER defined scissor
<---> [04.01] QP corrections report
<---> [01] Dipoles
<---> [DIP] Checking dipoles header
<---> [WARNING] [r,Vnl^pseudo] included in position and velocity dipoles.
<---> [WARNING] In case H contains other non local terms, these are neglected
<---> [02] Bethe Salpeter Equation @q1
<---> [02.01] Transition Groups build-up @q1
<02s> [BSK] Size (resonant): 40375
<02s> [BSK] (total): 40375
<02s> [BSK] Matricies : 1
<02s> [02.02] Independent Particles properties @q1
<02s> [DIP] Checking dipoles header
<02s> [02.03] BSE Kernel @q1 (Resonant CORRRELATION EXCHANGE)
<02s> Loading BSE kernel | | [000%] --(E <04s> Loading BSE kernel |########################################| [100%] --(E) --(X)
<04s> [03] BSE solver(s) @q1
<04s> [LA] SERIAL linear algebra
<04s> [03.01] Diago Solver @q1
<17s> BSK diagonalize | | [000%] --(E) --(X)[amir-System-Product-Name:103755] *** Process received signal ***
[amir-System-Product-Name:103755] Signal: Segmentation fault (11)
[amir-System-Product-Name:103755] Signal code: Address not mapped (1)
[amir-System-Product-Name:103755] Failing at address: 0x55eb4e09f8cc
[amir-System-Product-Name:103755] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x153c0)[0x7f69a024b3c0]
[amir-System-Product-Name:103755] [ 1] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cgezero+0x1c8)[0x7f69a08e8d78]
[amir-System-Product-Name:103755] [ 2] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cmmJIK2+0x71d)[0x7f69a090e7dd]
[amir-System-Product-Name:103755] [ 3] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cmmJIK+0x750)[0x7f69a090ef80]
[amir-System-Product-Name:103755] [ 4] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cgemmNC+0x229)[0x7f69a08db839]
[amir-System-Product-Name:103755] [ 5] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cher2kUN+0xec)[0x7f69a090abbc]
[amir-System-Product-Name:103755] [ 6] /lib/x86_64-linux-gnu/libblas.so.3(ATL_rher2kUN+0xae)[0x7f69a0a3ddbe]
[amir-System-Product-Name:103755] [ 7] /lib/x86_64-linux-gnu/libblas.so.3(ATL_cher2k+0x114)[0x7f69a090a714]
[amir-System-Product-Name:103755] [ 8] /lib/x86_64-linux-gnu/libblas.so.3(cher2k_+0x273)[0x7f69a0bc0023]
[amir-System-Product-Name:103755] [ 9] /lib/x86_64-linux-gnu/liblapack.so.3(chetrd_+0x768)[0x7f69a0d083c8]
[amir-System-Product-Name:103755] [10] /lib/x86_64-linux-gnu/liblapack.so.3(cheev_+0x304)[0x7f69a0cf6484]
[amir-System-Product-Name:103755] [11] /home/amir/yambo-5.0.3/bin/yambo(+0x244946)[0x55eb13a18946]
[amir-System-Product-Name:103755] [12] /home/amir/yambo-5.0.3/bin/yambo(+0x2416fa)[0x55eb13a156fa]
[amir-System-Product-Name:103755] [13] /home/amir/yambo-5.0.3/bin/yambo(+0x60819)[0x55eb13834819]
[amir-System-Product-Name:103755] [14] /home/amir/yambo-5.0.3/bin/yambo(+0x40f7f)[0x55eb13814f7f]
[amir-System-Product-Name:103755] [15] /home/amir/yambo-5.0.3/bin/yambo(+0x29a64)[0x55eb137fda64]
[amir-System-Product-Name:103755] [16] /home/amir/yambo-5.0.3/bin/yambo(+0x283fb)[0x55eb137fc3fb]
[amir-System-Product-Name:103755] [17] /home/amir/yambo-5.0.3/bin/yambo(+0x2c3cfc)[0x55eb13a97cfc]
[amir-System-Product-Name:103755] [18] /home/amir/yambo-5.0.3/bin/yambo(+0x2639a)[0x55eb137fa39a]
[amir-System-Product-Name:103755] [19] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3)[0x7f69a006b0b3]
[amir-System-Product-Name:103755] [20] /home/amir/yambo-5.0.3/bin/yambo(+0x2644e)[0x55eb137fa44e]
[amir-System-Product-Name:103755] *** End of error message ***
Segmentation fault (core dumped)
//

Please help me to solve this issue.

Best regards
Aref
Aref. Abeshipour
MSc student at chamran university of Ahvaz, Iran
https://scu.ac.ir/

User avatar
Daniele Varsano
Posts: 3809
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Error in BSE calculation :BSE solver: diagonalization

Post by Daniele Varsano » Sun Aug 01, 2021 5:47 pm

Dear Aref,

in order to spot the problem it would be better to have a loog to your input and report files: you can attach the files by renaming them with an allowed suffix e.g. .txt

In any case looking to your log file you have a rather large BSE matrix:

Code: Select all

 [BSK] Size (resonant): 40375
and it is not straightforward to diagonalise using serial linear algebra.

Here some suggestions and options:

1) Check if you really needs such large matrix (e.g. number of BSE bands I cannot tell more as the report is not provided).
2) use iterative approach e.g. Haydock if you do need eigenvectors
3) compile the code linking sacalapck libraries and use parallel linear algebra
4) Alternatively you can use the slepc libraries, if you need few eigenvectors and not the entire spectrum,

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/

Abeshipour
Posts: 11
Joined: Mon Mar 22, 2021 10:45 am

Re: Error in BSE calculation :BSE solver: diagonalization

Post by Abeshipour » Sun Aug 01, 2021 9:18 pm

Dear Danielle

Thanks for the helpful answer.

About your first recommendation, I attached the input files (bse_solve)
bse_solve.rar
. It would be great if you help me is it necessary to use this number of bands or not?

Best regards
Aref
You do not have the required permissions to view the files attached to this post.
Aref. Abeshipour
MSc student at chamran university of Ahvaz, Iran
https://scu.ac.ir/

User avatar
Daniele Varsano
Posts: 3809
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Error in BSE calculation :BSE solver: diagonalization

Post by Daniele Varsano » Mon Aug 02, 2021 7:31 am

Dear Aref,

the convergence with respect the number of bands depends it is system dependent and also depends on the range of energy you are interested.
I suggest you to start with few bands across the gap and smoothly increase it. It is possibile you need less bands than the number you are actually using.
If this is not the case and you really need such big matrix, then you can switch to iterative method to solve the BSE (e.g. haydock)

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