Yambo Community Forum

Hello!

I'm working on a set of 4 materials and I've been running yambo calculations on them. They are all very very similar to each other and I was able to run a G0W0 calculation for two of them without any problems. For some reason the last 2 are giving me some trouble related to dipoles and I'm not really sure where the problem comes from. I'm attaching the "l-" and "r-" outputs of the calculation, the r_setup from initialization, my yambo input and also the standard error output from the cluster. I'm I thought it was memory problem so I tried to increase it a little bit but no luck(also a compiled yambo with the memory profile flag). Some help/tips would be very appreciated!
Regards,

Dear Bruno,
can you check if you have some other error messages in other log files?

Daniele

Dear Bruno,
you are attaching the log file of the master_cpu.

Can you check the other logs to see if one of the cores is giving a more specific error message?

You can try with or Best,
D.

Dear Bruno

I notice that you are using the Terminator and this requires more memory
because all dipoles are calculated.

My advise is to try to generate the input file with the -V RL flag
and reduce the number of plane waves.

just type yambo -D
to see the number of plane waves in the wave-function for example if you have
...
WF G-vectors : 1465
....

you can regenerate your input file with -V RL flag and set a number of plane-waves up to 4 times smaller, in my case:

FFTGvecs= 400 RL # [FFT] Plane-waves

and see if this solve your problem.

You should perform some test on this value to see if does not change any result,
just do it on the optica in independent particle approximation yambo -o c -V RL

best
Claudio

Dear Daniele, Davide and Claudio,

I've checked the log files using the mentioned grep, but all of them finishes with this very same error. I've also tried what Claudio said but even inserting the flag and changing it to lower values of FFTGvecs I'm still getting the same error, I've tried to go very low until 5000 RL (there are 38735RL) but even then the error persists. But I'm not sure if the flag is working for me. After I change it on the input and run the calculation, should the r- output show the same value from the input on the line "WF G-vectors" ? Because doesn't matter the value that I choose on the input, the r- output is always showing 38735 RL.

Regards,

Dear Bruno

post again your r_ and l_ file with the small FFT cutoff.

Then you can try to compile Yambo with different IO, for example adding or removing the flag in the configure

--enable-hdf5-par-io

does it work if you run in serial with a small FFT cutoff?

best
Claudio

Dear Claudio,

I'm sending the r- and l- outputs both running in parallel and in serial as requested. Although the serial version also gave the error, now the dipole calculation percentage bar actually moved a little bit.
Regards,

Dear Bruno

your error is very strange, so please put all input files in the forum
DFT, pseudo, yambo inputs and the config.log file.

Then try to redo all calculations from scratch, remove the SAVE and restart from the p2y

in the while I will try to reproduce your bug

You can also update to the last version on github... there are few corrections, but nothing related
to dipoles as fas as I know.

best
Claudio

Dear Claudio,

Following your advice I restarted from scratch and generated the SAVE folder again, interestingly it appears that it's working. I have no idea what might have caused the problem, but it's possible that was some cluster instability during the first calculation since they are updating and expanding the machine. Thank you very much!

Best regards,