segmentation fault for G0W0

[JianjunYang]

Dear All,
I am trying to do the G0W0 correction on some specific K-points which are not included in the initial uniform k-grid. The calculation is for the fcc Si. Following the instruction given by http://www.yambo-code.org/doc/inputs/ypp_kk.php, the new k-grid including those specific K-points have been generated with the command ypp -k k, and I have obtained the new WAVEfunctions on the new k-grid with a non-scf cal with PWScf. Then run yambo setup with the command yambo -i -V kpt, and have the variable IkXLim set properly. Copy the SAVE/db.pp from the previous not-shifted grid calculation to the new SAVE folder, ready to run the QP correction on the specific K-points with the setup command "yambo -g n -p p -V gp" by selecting the interested k-points at the %QPkrange block. However, when I tried to run with the command "yambo -F yambo.in -J test", I came to the segmentation fault as shown below. I tried to run on different machines, but there is still the same error. Could anyone give me some suggestions to have the problem fixed? Thanks a lot! I could provide the input files later if necessary.

Jianjun Yang

###################
was unable to find any relevant network interfaces:

Module: OpenFabrics (openib)
Host: litai06.westgrid.ca

Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------

__ __ ______ ____ _____
/\ \ /\ \\ _ \ /"\_/`\/\ _`\ /\ __`\
\ `\`\\/"/ \ \L\ \/\ \ \ \L\ \ \ \/\ \
`\ `\ /" \ \ __ \ \ \__\ \ \ _ <" \ \ \ \
`\ \ \ \ \ \/\ \ \ \_/\ \ \ \L\ \ \ \_\ \
\ \_\ \ \_\ \_\ \_\\ \_\ \____/\ \_____\
\/_/ \/_/\/_/\/_/ \/_/\/___/ \/_____/

<---> [01] Files & I/O Directories
<---> [02] CORE Variables Setup
<---> [02.01] Unit cells
<---> [02.02] Symmetries
<---> [02.03] RL shells
<---> [02.04] K-grid lattice
<---> [02.05] Energies [ev] & Occupations
<---> [03] Transferred momenta grid
<---> [04] Bare local and non-local Exchange-Correlation
<---> [FFT-HF/Rho] Mesh size: 9 9 9
<---> [WF-HF/Rho loader] Wfs (re)loading |####################| [100%] --(E) --(X)
<---> EXS | | [000%] --(E) --(X)[litai06:13892] *** Process received signal ***
[litai06:13892] Signal: Segmentation fault (11)
[litai06:13892] Signal code: Address not mapped (1)
[litai06:13892] Failing at address: 0x350
[litai06:13892] [ 0] /lib64/libpthread.so.0 [0x2ace49b04b70]
[litai06:13892] [ 1] yambo(xco_hartree_fock_+0xe7a) [0x44dfba]
[litai06:13892] [ 2] yambo(xco_driver_+0x4d8) [0x44c188]
[litai06:13892] [ 3] yambo(yambo_driver_+0x28c) [0x41a33c]
[litai06:13892] [ 4] yambo(main+0xd06) [0x418b16]
[litai06:13892] [ 5] /lib64/libc.so.6(__libc_start_main+0xf4) [0x2ace49d2e9b4]
[litai06:13892] [ 6] yambo [0x417d59]
[litai06:13892] *** End of error message ***
Segmentation fault

[andrea marini]

Dear All,
Could anyone give me some suggestions to have the problem fixed? Thanks a lot! I could provide the input files later if necessary.

As far as I can see from your post it seems like you are following the correct procedure. Therefore I cannot really judge if it is a Yambo or a user problem. Please post all relevant input files and I will try to run the GPL version of the code and reproduce your error.

Thanks

Andrea

[JianjunYang]

Dear Andrea,
This is a very small cal on bulk Si. I am very eager to learn how to successfuly run the QP correction on the bandstructure.
Here attached the relevant input and/or output files.

Thank you very much!
Jianjun

Dear All,
Could anyone give me some suggestions to have the problem fixed? Thanks a lot! I could provide the input files later if necessary.

As far as I can see from your post it seems like you are following the correct procedure. Therefore I cannot really judge if it is a Yambo or a user problem. Please post all relevant input files and I will try to run the GPL version of the code and reproduce your error.

Thanks

Andrea

[andrea marini]

Dear JianjunYang,

I redid the calculation as you did with only a minor change. In the ypp I used

Code: Select all

cooOut= "iku"                # Points coordinates (out) cc/rlu/iku/alat

This is beacuse otherwise PWscf assumes the input is in 2pi/a. If you want to use rlu instead of iku you can try to run PW with "crystal" in the line of the input caontaining the "K_LATTICE" string.

In this way I did not get anyu seg-fault.

Andrea

[JianjunYang]

Dear Andrea,
Thank you so much for the help. It does work now.

One more question here. In principle, I could get the G0W0 correction on the entire bandstructure provided all necessary K-points are given in ypp file. Is there any other better way to have the job done?

Thanks again,
Jianjun

Dear JianjunYang,

I redid the calculation as you did with only a minor change. In the ypp I used

Code: Select all

cooOut= "iku"                # Points coordinates (out) cc/rlu/iku/alat

This is beacuse otherwise PWscf assumes the input is in 2pi/a. If you want to use rlu instead of iku you can try to run PW with "crystal" in the line of the input caontaining the "K_LATTICE" string.

In this way I did not get anyu seg-fault.

Andrea

[andrea marini]

Dear Andrea,
Thank you so much for the help. It does work now.

You're welcome

One more question here. In principle, I could get the G0W0 correction on the entire bandstructure provided all necessary K-points are given in ypp file. Is there any other better way to have the job done?

Yes. By calculating the corrections on a few points and then interpolating. This will be possible shortly using the interface between Yambo and Wannier90 that will be released in the near future.

Andrea

[JianjunYang]

Dear Andrea,
Thanks again. I will keep an eye on the released date of this interface.

Jianjun

[/quote]

Yes. By calculating the corrections on a few points and then interpolating. This will be possible shortly using the interface between Yambo and Wannier90 that will be released in the near future.

Andrea[/quote]