yambo 4.2.0 - NetCDF: Variable not found

[gyro555in]

Hi,

We recently installed Yambo 4.2.0 with IOTK, NetCDF ... support thanks to our IT administrator. I am running a tutorial on GW(Silicon) and with the given tutorial input file and I get the following error while running yambo -F INPUTS/01HF_corrections -J HF_7Ry

[ERROR] STOP signal received while in :[04] Bare local and non-local Exchange-Correlation
[ERROR] File ./SAVE//ns.wf_fragments_1_1; Variable WF_COMPONENTS_@_SP_POL1_K1_BAND_GRP_1; NetCDF: Variable not found

When You check this like ncdump -h SAVE/ns.wf_fragments_1_1. You can see that there are only the variables WF_REAL_COMPONENTS_@_K1_BAND_GRP_1 and WF_IM_COMPONENTS_@_K1_BAND_GRP_1 exisiting. Our IT admin says it is the fault with the Input file rather than the installation. Can someone point what's going on?

Thanks,
Kushal

[Daniele Varsano]

Dear Kushal,
please sign your posts with your complete affiliation: this is a rule of the forum. You can do that once for all by filling your signature in your profile.
In 4.2 the format of the wf database has changed and it looks you are using old databases.
You can either generate new database using p2y after a quantum espresso calculations, or convert the database from the old to the new format by using

Code: Select all

ypp -z 

Finally, at this link, you can find new yambo tutorials we prepared for the last school we had this year.

Best,

Daniele

[gyro555in]

Thanks Danielle for the prompt reply. Following the hBN tutorial, I converted to new format using ypp -z and it generated a new folder "FixSAVE" with the ns.db1, ns.wf, ns.wf_fragments_x_1...., ns.kb_pp_pwscf, ns.kb_pp_pwscf_fragment_x.... . I moved these files to the original SAVE and it is now reading these files but I get the following error.

---> [01] CPU structure, 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
<---> [WF] Performing Wave-Functions I/O from ./SAVE
<---> [FFT-HF/Rho] Mesh size: 3 3 12[hypnos5:15667] *** Process received signal ***
[hypnos5:15667] Signal: Segmentation fault (11)
[hypnos5:15667] Signal code: Address not mapped (1)
[hypnos5:15667] Failing at address: 0x7fff8fa1a10c
[hypnos5:15667] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10330)[0x7f330ca0e330]
[hypnos5:15667] [ 1] /usr/lib/libblas.so.3(ATL_cdotc_xp0yp0aXbX+0x20)[0x7f330da48600]
[hypnos5:15667] *** End of error message ***
Segmentation fault (core dumped)

[Daniele Varsano]

Dear Kushal,
can you please post the input file you are giving problems? We will try to reproduce the error.
Best,

Daniele

[gyro555in]

HF_and_locXC # [R XX] Hartree-Fock Self-energy and Vxc
FFTGvecs= 1 Ry # [FFT] Plane-waves
EXXRLvcs= 1 Ry # [XX] Exchange RL components
%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 14| 1|100|
%

[Daniele Varsano]

Dear Kushal,
I do not know if this is the problem, I have still not tested your input, but:
You are using a very small FFT, which will give you surely non physical results: the wave function are very badly represented:
Next:

Code: Select all

%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 14| 1|100|
%

you are asking to calculate a lot of energy corrections (1400!). Usually one is interested in bands around the ban gap.
Such a big number can cause you run out of memory.
Can you repeat the calculation using more reasonable number? Leave the FFTGvecs as default and set QPkrange something as:
1|14|7|10
or alike.

Best,
Daniele

[gyro555in]

The error persists inspite of changing the suggested parameters. I tried reducing more but still get the same error.

[Daniele Varsano]

Dar Kushal,
thanks, we will have a look and let you know.

Daniele

[Davide Sangalli]

Dear Kushal,
a stupid comment, just to be sure.

Once you convert the WFs to the new format, you should just use the "new" SAVE, not copy the files from the folder FixSave/SAVE in the "old" SAVE.
If you do the latter procedure the SAVE folder may contain a mixture of new an old files and that may cause problems.

Best,
D.

[gyro555in]

Of course, that's what I meant to imply. Is there any other simple input to test this new version?