MadGraph5_aMC@NLO

Pythia Error in Multicore

Asked by Joel Jones

Dear MadGraph,

I have an issue when performing the shower with more than one core. I guess this is a Pythia issue, not a MadGraph one, but at the same time I understand that the multicore execution of Pythia is actually a feature of MadGraph.

Anyway, I updated MadGraph (v 3.4.0) and Pythia (v 8306), and now I cannot perform a multicore shower (one core does work). I get the following error.

***
INFO: Running Pythia8 [arXiv:1410.3012]
Splitting .lhe event file for PY8 parallelization...
Submitting Pythia8 jobs...
Pythia8 shower jobs: 1 Idle, 3 Running, 0 Done [2 seconds]
WARNING: program /home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/Events/run_02/PY8_parallelization/split_0/run_PY8.sh 0 launch ends with non zero status: 127. Stop all computation
Pythia8 shower jobs: 0 Idle, 3 Running, 1 Done [6 seconds]
Pythia8 shower jobs: 0 Idle, 2 Running, 2 Done [6 seconds]
Terminated
Pythia8 shower jobs: 0 Idle, 1 Running, 3 Done [6 seconds]
Pythia8 shower jobs: 0 Idle, 0 Running, 4 Done [6 seconds]
Command "generate_events " interrupted with error:
Exception : program /home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/Events/run_02/PY8_parallelization/split_0/run_PY8.sh 0 launch ends with non zero status: 127. Stop all computation
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/run_02_tag_1_debug.log'.
Please attach this file to your report.
***

Even though I have other versions of Pythia installed (on old MadGraph copies), I have checked that the paths on the PYTHIA8 and PYTHIA8DATA are consistent:
echo $PYTHIA8
/home/jjones/Programs/MG5_aMC_v3_4_0/HEPTools/pythia8
echo $PYTHIA8DATA
/home/jjones/Programs/MG5_aMC_v3_4_0/HEPTools/pythia8/share/Pythia8/xmldoc

Any suggestions are appreciated!!!

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
Last query:
Last reply:
Revision history for this message
Joel Jones (n-jones-j) said : 		#1

Info from run_02_tag_1_debug.log
**************
generate_events
Traceback (most recent call last):
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1544, in onecmd
    return self.onecmd_orig(line, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1493, in onecmd_orig
    return func(arg, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/madevent_interface.py", line 2404, in do_generate_events
    self.run_generate_events(switch_mode, args)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/common_run_interface.py", line 7613, in new_fct
    original_fct(obj, *args, **opts)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/madevent_interface.py", line 2669, in run_generate_events
    self.exec_cmd('shower --no_default', postcmd=False, printcmd=False)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1573, in exec_cmd
    stop = Cmd.onecmd_orig(current_interface, line, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1493, in onecmd_orig
    return func(arg, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/madevent_interface.py", line 4004, in do_shower
    self.exec_cmd('%s %s'%(shower,' '.join(args)),
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1573, in exec_cmd
    stop = Cmd.onecmd_orig(current_interface, line, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/extended_cmd.py", line 1493, in onecmd_orig
    return func(arg, **opt)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/madevent_interface.py", line 4718, in do_pythia8
    self.cluster.wait(parallelization_dir,wait_monitoring)
  File "/home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/bin/internal/cluster.py", line 829, in wait
    raise Exception(self.fail_msg)
Exception: program /home/jjones/Programs/MG5_aMC_v3_4_0/TTtest/Events/run_02/PY8_parallelization/split_0/run_PY8.sh 0 launch ends with non zero status: 127. Stop all computation

Revision history for this message
Joel Jones (n-jones-j) said : 		#2

In addition, the PY8_log.txt in PY8_parallelization/split_0/ directory ends with this message:

./MG5aMC_PY8_interface: symbol lookup error: ./MG5aMC_PY8_interface: undefined symbol: _ZN5HepMC15WeightContainer9push_backERKdNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#3

So I guess that you need to re-install
MG5aMC_PY8_interface
since it seems to have been compile with an older compiler/library or something in that spirit and therefore fails to link to your new pythia8 version.

Cheers,

Olivier

Revision history for this message
Joel Jones (n-jones-j) said (last edit ): 		#4

Mmmm, I did reinstall MG5aMC_PY8_interface, (and I did it once more just in case), and the problem persists.

Note that the installation gives me the following warning:

/home/jjones/Programs/MG5_aMC_v3_4_0/HEPTools/HEPToolsInstallers/HEPToolInstaller.py:15: DeprecationWarning: The distutils package is deprecated and slated for removal in Python 3.12. Use setuptools or check PEP 632 for potential alternatives from distutils.version import LooseVersion, StrictVersion

Revision history for this message
Best Olivier Mattelaer (olivier-mattelaer) said : 		#5

Then maybe an issue with your LD_LIBRARY_PATH ?

Thanks for the notification for the HEPToolsInstallers, I will handle that when the first release candidate of python3.12 will be available (so in approximativelly a year)

Revision history for this message
Joel Jones (n-jones-j) said : 		#6

Ok, regarding LD_LIBRARY_PATH, my path to fastjet was outdated, but after correcting it, reinstalling Pythia, and reinstalling MG5aMC_PY8_interface, I still find the same problem.

I have never included any Pythia info on my LD_LIBRARY_PATH, though. I added the directories within MadGraph, but still does not work. :-/

Revision history for this message
Joel Jones (n-jones-j) said : 		#8

Ok, this took a while, but it is now solved.

It was LD_LIBRARY_PATH, indeed. I was including a relatively old hepmc installation on the path, and for some reason this broke everything. After removing it, things work again. Many thanks!

Revision history for this message
Joel Jones (n-jones-j) said : 		#9

Thanks Olivier Mattelaer, that solved my question.

Revision history for this message
Aviral (elf-legolas) said (last edit ): 		#10

Hello Joel,
I am facing a similar problem. Could you mention how did you amend your LD_LIBRARY_PATH?

Revision history for this message
Joel Jones (n-jones-j) said : 		#11

Oh dear, this was too long ago, I don't remember exactly. What must have happened was the following: I had some installation of the HepMC library on my system (https://gitlab.cern.ch/hepmc/HepMC3), and even though I was not using it, I had it included in the LD_LIBRARY_PATH. So I just removed all mention of hepmc there.