MADGRAPH HEPMC output

[dirkzerwas]

MADGRAPH HEPMC output not yet implemented
Need to add to the python suite hadronization and writing via pythia

[dirkzerwas]

I am wondering if the installation of madgraph in KEY4HEP is complete:
I added
install pythia8
as well as
install mg5amc_py8_interface
to the madgraph input file, both commands are recognised:
"The import format was not given, so we guess it as command
install pythia8
You are installing 'pythia8', please cite ref(s): arXiv:1410.3012.
"
but result in an error:
OSError : [Errno 30] Read-only file system: /cvmfs/sw-nightlies.hsf.org/key4hep/releases/2024-03-08/x86_64-centos7-gcc12.2.0-opt/madgraph5amc/3.4.2-jzrczx/HEPTools'

in the directory .....3.4.2-jzrczx there is no HEPTools executable installed.

[Zehvogel]

I am wondering if the installation of madgraph in KEY4HEP is complete:

Looking at the maintenance state of the spack package I would not be surprised if stuff is missing... maybe @jmcarcell can comment :)

[dirkzerwas]

To add some more info, here are the steps:

1. set up KEY4HEP:
   source /cvmfs/sw-nightlies.hsf.org/key4hep/setup.sh
2. launch madgraph:
   mg5_aMC
3. try to "install pythia8" gives this message:

MG5_aMC>install pythia8
You are installing 'pythia8', please cite ref(s): arXiv:1410.3012.
Command "install pythia8" interrupted with error:
OSError : [Errno 30] Read-only file system: '/cvmfs/sw-nightlies.hsf.org/key4hep/releases/2024-03-08/x86_64-centos7-gcc12.2.0-opt/madgraph5amc/3.4.2-jzrczx/HEPTools'
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in 'MG5_debug'.
Please attach this file to your report.

I checked in /cvmfs/sw-nightlies.hsf.org/key4hep/releases/2024-03-08/x86_64-centos7-gcc12.2.0-opt/madgraph5amc/3.4.2-jzrczx/
there is no directory "HEPTools"

from the madgraph documentation I think that "install BLA" first downloads a tool "HEPToolsInstaller" probably into the mg5 directory in order to install pythi8, mg5_pythia8, so I guess that's what's blocking the execution.

BTW in
/cvmfs/sw-nightlies.hsf.org/key4hep/releases/2024-03-08/x86_64-centos7-gcc12.2.0-opt/madgraph5amc/3.4.2-jzrczx/vendor/
the tool seems to be there, but not installed.
OfflineHEPToolsInstaller.tar.gz

I guess the tool then installs the interface and probably some code when calling "install mg5amc_py8_interface"
for "install pythia8" this may simply be the setting of the path. pythia8_path = ../pythia8 which we could try to work around (not sure).

Cheers,
Dirk

[dirkzerwas]

Result of temporary addition of LHE read:
Difermion_Polarized:
Madgraph 4.0503 +- 0.089162 pb

Muon:
Sherpa 1719.95 +- 1.07523 pb
Madgraph 1720.44 +- 0.450408 pb
Whizard 1721.66 +- 2.39159 pb

MuonNeutrino:
Madgraph 3949.8 +- 0.48569 pb
Whizard 3949.96 +- 0.465219 pb
Sherpa 3948.88 +- 0.0907291 pb

Tau:
Whizard 2005.99 +- 0.219619 pb
Sherpa 2010.39 +- 0.0450829 pb
Madgraph 2006.9 +- 0.536246 pb

TauNeutrino:
Sherpa 3948.94 +- 0.0900616 pb
Whizard 3949.9 +- 0.456594 pb
Madgraph 3950.2 +- 0.46724 pb

ZH:
Madgraph 0.24017 +- 0.00012676 pb
Whizard 0.239985 +- 4.45033e-06 pb
Sherpa 0.252791 +- 1.94466e-05 pb

ZHDecay:
Madgraph 1.72148e-06 +- 1.49455e-09 pb
Whizard 0.238792 +- 0.000112203 pb

ZHDecay: I think MADGRAPH gives out sigma(ZH)*BR(H)*BR(Z) whereas Whizard sticks to sigma(ZH).
Need to check if there is a way get this consistently

BTW @apricePhy do you know if the lhef files of MADGRAPH are "standard"? The HepMC LHEF-Reader went into an infinite loop. I cured it by removing the content between

and (containing the Madgraph cards summary) from the file via sed. I did not check if it was a subsection <></> or the full header. It seemed like standard xml marking....

[dirkzerwas]

Hi,
a couple more things about the MG-pythia interface:

• install pythia8 installs both PYTHIA8 but more importantly also mg5amc_py8_interface
• one can probably get around the installation of PYTHIA8 and HEPMC by defining in the madgraph configuration file:
• pythia8_path = ABSOLUTEPATH2PYTHIAINKEY4HEP
• hepmc is more delicate, since pythia8 is probably still at HEPMC2, but this might work
• hepmc_path = ABSOLUTEPATH2.HEPMCINKEY4HEP

However the driver of the pythia8 interface resides in mg5, not in PYTHIA, therefore the absence of the directory:
HEPTOOLS (and its subdirectories) prevents the

• install mg5amc_py8_interface
  command from running (even assuming that setting the paths of pythia8 and hepmc will be sufficient)

My guess is that during installation one would have to do the following:

• git clone or untar mg5
• update the paths in mg5/input/mg5_configuration.txt
• execute:
  bin/mg5_aMC INPUT.dat
  where input contains the line:
  install mg5amc_py8_interface

This should install the interface in the release directorymg5/HEPTools/
when I tried it locally:
install mg5amc_py8_interface
also did the compilation so in
directorymg5/HEPTools/MG5aMC_PY8_interface
the "pythia8 driver/interface" executable:
MG5aMC_PY8_interface
is available

[dirkzerwas]

executing:
mg5_aMC configureMadgraph

with configureMadgraph containing the line
install mg5amc_py8_interface

installs the interface to PYTHIA8 (drivers and all that).

Things that can go wrong:
-> in configureMadgraph you might need to as first line (before install...):
set pythia8_path XXXX
where XXX is the pythia8 directory in KEY4HEP.

[dirkzerwas]

Supplementary information:
the HepMC ReaderLHE goes into an infinite loop when reading the madgraph lhe file. Bug reported to Andrii and ANdy Buckley of HepMC. Andrii corrected the bug (ignore a comment anchor <!--). Update foreseen for HepMC 3.3. afaict

[dirkzerwas]

Update from Juan:

• for OS=ALMA there is a MG5 v 3.5.4 implementation including the installation of PYTHIA

However this does not work:

• the command "output OUTPUT" leads to an error in the python script various/misc.py when the file attributes are copied from the Template directory to the local directory.

Further checks:

• the problem is neither related to the MG version nor the pythia interface installation:
• Running OS=ALMA; MG5 v 3.4.2 (which does not have Pythia installed) has the same error message
• Running OS=centos: MG5 v3.4.2 (which also does not have Pythia installed) works out of the box

The error seems to be OS related. I asked Juan if MG5 v3.5.4 WITH Pythia can be installed on centos for a cross check.