Yambo Community Forum

Hi,
I'm using Yambo-4.3.3 to run real-time calculations. The test on silicon was passed.

However, for my own system, the results for real-time BSE are weird.
1) The external electrical field is set in the direction of [1 1 1]. The real-time IPA calculations have non-zero initial polarizations for three directions. After post-processing, the dielectric functions are nicely agreement with those of frequency-domain IPA calculations.
2) However, for real-time BSE calculations, the initial polarization along the direction X is zero. And subsequent polarizations are always zero along direction X but not zero in the directions of y and z.

Some necessary files were attached, expect any suggestions!

Thanks a lot!

Best,
Xiaowei

Hallo Xiaowei

I could not spot anything wrong with what you sent. Can you post as well the r-03_lr_test_nloptics?

Also, the run you posted for BSE did not finish. Did you stop it or did it finish with an error? In case, what was the error?

When you say that the Silicon test was passed are you running the test suite for yambo_nl or was it your own testrun? I am asking it since silicon in the test-suite does not contain a test for yambo_nl. The system you are running is too big for debugging so it would be helpful to have a smaller system where the problem occurs.

One thing to check is whether the problem occurs only with the human readable file or also with the database.
you can use the ncdump executable (if you compile the netcdf from scratch it is in the yambo bin, otherwise it should be available as shell command) and check the ndb.Nonlinear_fragment_1 (which in your case should be in the 03_lr_tests directory, so YAMBO_BIN_DIR/ncdump 03_lr_tests/ndb.Nonlinear_fragment_1) and look at NL_P_freq_0001 (are there zeros as well?)

Best,
Myrta

Dear Prof. Grüning,
Sorry for the delay. It took me some time to finish the real-time calculations.

[/quote]

myrta gruning wrote:Can you post as well the r-03_lr_test_nloptics?

1) Yes, I've uploaed it.

myrta gruning wrote:Also, the run you posted for BSE did not finish. Did you stop it or did it finish with an error? In case, what was the error?

2)I stoped it. These days, I finished the run and there was no error. Please see the attached file "yambo-4.3.3-new"

myrta gruning wrote:When you say that the Silicon test was passed are you running the test suite for yambo_nl or was it your own testrun? I am asking it since silicon in the test-suite does not contain a test for yambo_nl. The system you are running is too big for debugging so it would be helpful to have a smaller system where the problem occurs.

3)I did the Silicon test by following the tutorial on Lumen's website. The results are reasonable for ipa and bse. I didn't come across this problem for other systems so far, but I can try more systems.

myrta gruning wrote: look at NL_P_freq_0001 (are there zeros as well?)

4)I've uploaded the data. There are a lot of zeros at the beginning.

Thanks
Best,
Xiaowei

Dear Prof. Grüning,
When I read the soure code of Yambo-4.3.3, I found a problem.
For these lines in nloptics/Nl_start_and_restart.F,
<<< if(l_use_DIPOLES) then
call DIP_polarization(E,k,V_bands,NL_initial_P)
else
! This subroutine update also the Sm1_plus,Sm1_minus and All_S_det matrices
call Berry_Polarization_NEQ(E,k,V_bands,NL_initial_P)
!
endif>>>
the subroutine "Berry_Polarization_NEQ" doesn't exist in the whole yambo directory.
Maybe this is helpful.
Best,
Xiaowei

Hallo Xiaowei
I'll look at the tar file you post later. For the moment, Berry_Polarization_NEQ, should be in src/hamiltonian.

Note that the code would not compile if the source of a subroutine that is called in the code is not available.

Best,
Myrta

Dear Xiaowei

please try to repeat your calculations with the last version 4.4.
Different bugs were corrected in this version

https://github.com/yambo-code/yambo/arc ... 4.0.tar.gz

I will try to repeat your calculations and see what is wrong

best
Claudio

May you send us also the DFT input files
so we can reproduce your results

best regards
Claudio

Dear Claudio,

OK! I've tried the version 4.4. But the output polarizations are "NaN" for silicon and my system. I've collected all the related files and send to you.
Cause the tar file is big, I will send it to your email!

Thanks a lot!

Best
Xiaowei