Yambo Community Forum

Posted: **Sun Jun 26, 2022 8:23 pm**

Dear all,

I'm calculating quasiparticle corrections to DFT band structure for a 2D semiconductor and I'm using Yambo 5.1.
At the moment I'm studying convergence of the gap at Γ with respect to BndsRnXp and NGsBlkXp, but before that I'm trying to do a test calculation with the input gw_ppa.in here attached. The problem I run into is that the calculation stops at some point during the Dyson equation (I'm using Newton solver) with no errors in the report or in the log. I'm using slurm to run the calculation and the only errors I find are in the job.out file. This is how I run the job

Code: Select all

#! /bin/bash

#SBATCH --ntasks-per-node=16
#SBATCH --cpus-per-task=1
#SBATCH -o gw_ppa_jobcm.out
#SBATCH --job-name='gw_ppa_test'

mpirun -np $SLURM_NTASKS ~/Software/yambo-5.1.0/bin/yambo -F gw_ppa.in -J gw_ppa_test

and the errors in the job.out are

=================================================================================
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= RANK 15 PID 1411080 RUNNING AT cm01
= KILLED BY SIGNAL: 9 (Killed)
=================================================================================

Any help for this issue would be really appreciated. Thank you,

Matteo

Posted: **Tue Jun 28, 2022 11:08 am**

Dear developers,

I ran several tests changing parallelisation parameters and even on other machines because I thought it was a memory issue, but nothing changed. Eventually I removed the RIM_W method keeping the rest of the input unchanged and the calculation runs without any problem.
The relevant files are attached.

Thank you for your attention,

Matteo

Posted: **Wed Jun 29, 2022 10:59 am**

Dear all,
I met the same problem and I think it is not the memory problem at least.

shudong

Posted: **Mon Jul 04, 2022 9:24 am**

Dear Matteo and Shudong,
thanks for reporting, we will try to reproduce and spot the problem.
Best,
Daniele

Posted: **Mon Jul 04, 2022 3:47 pm**

Dear Matteo and Shudong,
thank you for your reports.

We find a bug in the RIM-W implementation in the 5.1 version of the code.
The bug has been fixed and will be available in the next patch of the code.

You have two possibilities: waiting for the new patch to be released, or to solve the bug yourself in the
version 5.0. It is very easy and explained in the following

-------------------
In the file src/qp/QP_interpolate_W.F, move the line 283:

YAMBO_FREE(f_coeff)

between the line 360 and 361 (just before the end of the subroutine).
-------------------

Best,
Alberto

Posted: **Mon Jul 04, 2022 5:22 pm**

Thank you Alberto, I'm glad it has been solved

Best,

Matteo

Posted: **Tue Jul 05, 2022 2:51 am**

Dear Daniele and Alberto,
I have fixed the problem as suggested by Alberto, and now it works!
Thanks for your kindly help!

Best,

Shudong

Yambo Community Forum

Calculation stopping with "BAD TERMINATION" -- RIM-W

Calculation stopping with "BAD TERMINATION" -- RIM-W

Calculation stopping with "BAD TERMINATION" -- RIM-W

Re: Calculation stopping with "BAD TERMINATION" -- RIM-W

Re: Calculation stopping with "BAD TERMINATION" -- RIM-W

Re: Calculation stopping with "BAD TERMINATION" -- RIM-W

Re: Calculation stopping with "BAD TERMINATION" -- RIM-W

Re: Calculation stopping with "BAD TERMINATION" -- RIM-W