Sigma Calculation ERROR

[Ritwik Das]

Dear BerkeleyGW users and developers, 

I am new to BGW. I compiled the BGW v_3.0.1 on Redhat Linux using Intel OneAPI (v_2022.1.2.146) libraries in parallel. Everything was compiled successfully without any error. My arch.mk is in the attachment. 

I am using QuantumEspresso v_7.0 for mean-field calculations. I am running calculations in the "6a-SOC" directory from BerkeleyGW Workshop-BESC2022 tutorials (BerkeleyGW_Tutorial_2022.tar.gz - link to download). All mean-field and epsilon calculations finished successfully but the sigma calculation (with both sigma.cplx.x and sigma.real.x) failed with the following error: 

"forrtl: severe (174): SIGSEGV, segmentation fault occurred" 

Please find the attached for sigma.out file for scalar and full relativistic calculations. (The sigma.log file is empty.) 

[I changed inputs accordingly in the mf folder to generate WFN and RHO files for sigma.real.x calculation. Again received the same error. Tried to compile the BGW code with different configurations in arch.mk but nothing helped.] 

Then for a sanity check, I run the testsuite in parallel. But 14/27 tests failed (used $ make check-parallel inside testsuite directory). The whole test output log (BGW-test.log) is in the attachment. The summary of the test run is below: 

    testfile                                          # failed testcases
    --------------------------------------------------------------------
    AlAs-DFT/AlAs.test                                1
    BN/BN.test                                        1
    GaAs-EPM/GaAs_spinor.test                         1
    GaAs-EPM/GaAs.test                                1
    Graphene/Graphene_3D.test                         1
    Graphene/Graphene.test                            1
    Si-EPM/Si_cplx_spin.test                          1
    Si-EPM/Si_cplx.test                               1
    Si-EPM/Si.test                                    1
    Si-EPM_subspace/Si_subspace_cplx_spin.test        1
    Si-EPM_subspace/Si_subspace_cplx.test             1
    Si-EPM_subspace/Si_subspace.test                  1
    Silane-PARATEC/Silane.test                        1
    Si-Wire-EPM/Si_wire.test                          1

Could someone please help me to resolve the issue? 

Thank you very much in advance for your consideration. 

Best regards, 

Ritwik DAS 

Centre for Nanoscience and Nanotechnology (C2N)
C2N UMR CNRS 9001 - Université Paris-Saclay
Department: Physics (Nanoelectronics)
10 Boulevard Thomas Gobert, 91120 Palaiseau, France 

[Brad Barker]

Hi Ritwik,

When you compiled, did you generate both "real" and "complex" flavors for all executables? The first thing that comes to mind seeing the list of failed test cases is that they would require complex executables.

Best,

Brad

--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/0a674b77-bac9-4c40-9d15-ea745edba457n%40berkeleygw.org.

[Ritwik Das]

Hello Brad, 

Thank you for your response. Yes, I generated both "real" and "complex" flavours for all executables using "make -j 8 all-flavors". 

If you look at the output log of the sigma calculation ("sigma.out - full_rel" file in the previous email), you will find what makes the program stop. 

sigma.cplx.x       00000000004DBBAA  inread_m_mp_inrea          44    inread.f90

sigma.cplx.x       00000000004C63C9  input_m_mp_input_         100    input.f90
sigma.cplx.x       000000000052860D  MAIN__                              238    sigma_main.f90

I think, this might be associated with the Intel compiler (some file linker breaks, maybe?). You'll find a similar issue posted here: https://groups.google.com/a/berkeleygw.org/g/help/c/KupAgLF8KCI/m/-w7iuf6hAwAJ. But the provided solution does not work for me. 

I am also using the most recent Intel OneAPI v2022. 

Thanks. 

Best, 

Ritwik 

[Mauro Del Ben]

Hi Ritwik

Can you try to run this command before running?

ulimit -s unlimited

Best

-M

To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/5edb2901-6dff-47b4-9997-8f8fd30f8913n%40berkeleygw.org.

[Ritwik Das]

Hello Mauro, 

Thank you very much. It indeed worked. 

So, it was a memory problem. And the command "ulimit -s unlimited" sets the system dump file size to unlimited. 

Am I right? 

Ritwik 

[Mauro Del Ben]

Hi Ritwik,

Good to know the problem was fixed. 

It is not a problem of memory per se, it is just the stack size limit set on your system. Some of our executables used preallocated quantities that might overflow the stack limit, that's why setting it to unlimited fixes the problem. In the next release we have moved to dynamical allocation of such arrays and the stack limit will no longer be an issue.

Best

-M

[Ritwik Das]

Hi Mauro, 

That's great. Thanks for letting us know! 

-- Ritwik

[Wenqi Yang]

Dear BWG users and developers,

I also meet the same problem. However, ever after I run this command, ulimit -s unlimited, I get the same error. Could you please give me some other suggestions? Thank you very much!

Best

Wenqi

[Mauro Del Ben]

Hi Wenqi,

Can you give us more details? How do you compile? How are you running? 

Best

-M