Yambo Community Forum

Dear Yambo Team,
I got empty fort.0 and no output file (nor ndb.gopts and ndb.kindx) from initialization of yambo (using yambo.in with 4000 RL) after successful p2y.
The case is Eu doped ZrO2 surface + 1 oxygen vacans + 1 molecule O2 on top of it.
The inputs for scf and nscf contain SOC with the constrained magnetization, as below.
The case without oxygen molecule above this surface, with similar setup, went well through the BSE.
What could be the problem?
With best regards,
Gosia
=============== part of scf input ================
ibrav=6, a=7.4828, c=60.0, nat=73, ntyp= 4,
ecutwfc = 80, nbnd=1000,
tot_charge=1.0, starting_charge(1)=3.0,
force_symmorphic =.true.
noncolin=.true., lspinorb=.true.,
constrained_magnetization = 'total',
fixed_magnetization(1) = 0.0
fixed_magnetization(2) = 0.0
fixed_magnetization(3) = 4.0
======================

Dear Gosia,
I do not think that the issue is related with your system.
This seems more a permission issue related to the filesystem or I/O problem related with compilation.
Is it possible that you end-up with a problematic node?
Can you try to run the code interactively and see if the error is reproduced?

Best,
Daniele

Dear Gosia,
could you send me the exact sequence of commands that you are running.
And also a list of file present in your directory.

Best,
Nicola

Dear Daniele and Nicola,
we tried various combinations and the error is reproducible.
I use QE and Yambo with Hubbard U.
I did p2y and got SAVE using yambo with Hubbard correction (that works for other cases).
Then I do:

module --force unload LUMI
export EBU_USER_PREFIX=/project/project_465001703/EasyBuild
module load LUMI/24.03 partition/C
module load Yambo/5.2.3-cpeGNU-24.03

JID=${SLURM_JOB_NAME#*_}-${SLURM_JOB_ID}
echo "JOB: $JID"

srun strace yambo -F yambo.in -C log-${JID}/

During reading from SAVE/ns.db1 I get some error.
File fort.0 is empty and contains only sign of new line "\n".
While slurm-R3socO1-s4-13611632.out is attached.

Thank you so much! This is really a strange error!

Best regards,
Gosia

Dear Gosia,
after the launch of p2y, have you launched also yambo without any argument? This is the initialization phase necessary before every run.
See this tutorial: https://wiki.yambo-code.eu/wiki/index.p ... properties

Could you send me the file yambo.in?

Best,
Nicola

Dear Nicola,
yes, I always run yambo with yambo.in in the same form, that is very short, as below.
I only change the number of RL for different cases.
This is the step which failed after p2y.

Thank you very much.
Gosia
============== yambo.in (2 lines after ### )=========
setup # [R INI] Initialization
MaxGvecs= 4000 RL # [INI] Max number of G-vectors planned to use
============================ end =================

Hi again,
I repeated the calculations with PZ functional instead of PBE and I did it without constrained magnetization.
The error is the same.

This example is different from everything else that I calculated with Yambo because it contains
a doped crystal surface and O2 molecule in the distance of around 5 Angstroem.
The doped surface without O2 molecule is going OK.

Is any reason that two disconnected systems (two densities separated by almost no density)
cannot be read by Yambo?

Best regards,
Gosia

I attach the scf.in and nscf.in files tarred in directory.

Dear Gosia,

I'm pretty sure the error in not related to the particular system you are dealing with.
This kind of error can happen if something went wrong in the generation of the SAVE directory, miscompilation or other I/O reason.

From your previous message, I see you are using DFT+U. By default, this is not allowed by Yambo even if it can be tricked, is this the case?

Can you try to run Yambo on a system calculated without the Hubbard U term, just to understand if this is the reason of the failure?

Best,

Daniele

Dear Daniele,

I am sorry to bother you again but no way to solve it in a simple way.
All experiments: +U vs no U, GGA vs LDA, crystal surface + molecules or only 2 molecules, various symbols for atomic oxygen with the same PP (O1, O2, O)
lead to the conclusion that whatever I do everything works except this specific case, namely ZrO2 doped with Eu + O2 above the surface.
Strange thing is that the code breaks with this density from the DFT for all setups started from scratch, and never breaks for the surface without molecule
or two molecules in the box without the surface.

May I ask you for some notes about this part of the code, some definitions what is calculated in yambo initialization of ndb.* files and in which subroutine?
Is there any zero in the denominator or something like this?

With best regards,
Gosia