Bug in MCParticle list in conversion from HepMC to LCIO through DDSIM

[eleogran]

I have a HepMC file with the correct MCParticle list.
When giving it as input to ddsim, the LCIO output file results in a wrong MCParticle list.

• HepMC file with 100k events e+e- -> Z -> nu N, with N -> eqq generated with DelphesPythia8:
  /afs/cern.ch/work/e/eleogran/public/MCParticleList_issue/HNL.hepmc
• event 198: printout of the list of particle pdg and vertex position from the hepmc file:
  /afs/cern.ch/work/e/eleogran/public/MCParticleList_issue/evt_198_mc_info.txt
• ddsim steering file:
  /afs/cern.ch/work/e/eleogran/public/MCParticleList_issue/fcc_steer.py
• LCIO output file:
  /afs/cern.ch/work/e/eleogran/public/MCParticleList_issue/HNLOutput.slcio
  NB: event 198 is event 2 in this file, i.e. to look at it: dumpevent HNLOutput.slcio 2 | less

In the MCParticle list from the lcio file, particles with index = 7 and > 44 have vertex position:
(-8.35e-01, 3.80e-01,-7.65e-02)
rather than
(-3.73e+02, 1.64e+02,-3.31e+01)

I suspect a problem in the Lorentz boost for the crossing angle, which in my steering file is set:
SIM.crossingAngleBoost = 0.015

If I change this to 0.0, particles with index = 7 and > 44 have vertex position:
(-6.54e-309, 2.88e-309,-5.80e-310)
while the other particles are unchanged.

I have used /cvmfs/clicdp.cern.ch/iLCSoft/builds/2020-02-07/x86_64-slc6-gcc62-opt/.
and run:
$ ddsim --steeringFile fcc_steer.py

[andresailer]

As far as I can tell neither setting of Lorentz boost gives you a correct result. With the lorentzboost you get a short lifetime for the N (right handed neutrino?), without the boost there is 0 lifetime. So I think the problem isn't in the Lorentzboost, or at least not exclusively.
The particles between 8 and 44 are not given to geant4 (quarks and "strings" and such), so their information isn't updated.

[eleogran]

Correct, neither result is the right one.
As the boost was affecting only the particles with the wrong vertex, I thought that there could be something weird in the boost. Do I understand from your comment that it's a problem of Geant4 then?

[andresailer]

I can't make that call yet. I don't think this is a Geant4 issue, more likely a HepMC and DD4hep issue.

[petricm]

Is there an easy way to convert hepmc to stdhep and we try with that?

[andresailer]

There is an issue with the creation of G4PrimaryParticles
in

DD4hep/DDG4/src/Geant4InputHandling.cpp

Line 354 in b530769

g4->SetMass(p->mass);

I added g4->Print() before and after this line, which gives:

==== PDGcode 9900012  Particle name nu_Re
 Assigned charge : 0
     Momentum ( -28.3765[GeV/c], 12.5034[GeV/c], -2.5186[GeV/c] )
     kinetic Energy : 31.0612 [GeV]
     Mass : 0.05 [GeV]
     Polarization ( 0, 0, 0 )
     Weight : 1
<<<< End of link
Massive
==== PDGcode 9900012  Particle name nu_Re
 Assigned charge : 0
     Momentum ( -58.1954[GeV/c], 25.6423[GeV/c], -5.16523[GeV/c] )
     kinetic Energy : 31.0612 [GeV]
     Mass : 50 [GeV]
     Polarization ( 0, 0, 0 )
     Weight : 1
<<<< End of link

EDIT: Part of the problem is addressed in #753 , in addition there is an inconsistency of the mass in particle.tbl and what is in the hepmc file.

[andresailer]

@eleogran Could you please try what happens when you set the mass for 9900012 in the particle.tbl file to 50 GeV?

[eleogran]

@andresailer the particle vertex is yet different, but still not correct.
Please find the file here: /afs/cern.ch/work/e/eleogran/public/MCParticleList_issue/evt198_fixedMN.slcio
At the same path the updated particle.tbl

[andresailer]

Thanks!
The hepmc2 reader does not treat the vertex information correctly.
This works with the HepMC3 reader. I'll try to find an iLCSoft installation with that enabled.
With that reader enabled I get

[00000020]    5|   9900012|-2.84e+01, 1.25e+01,-2.52e+00|-2.84e+01, 1.25e+01,-2.52e+00| 5.89e+01| 22 |[    c   ]| 0.00e+00, 0.00e+00, 0.00e+00|-3.73e+02, 1.64e+02,-3.31e+01| 5.00e+01| 0.00e+00| 0.00e+00, 0.00e+00, 0.00e+00|  (0, 0)   | [4] - [7,8,9] 

[eleogran]

Great! Thank you very much @andresailer

[eleogran]

Hello,
is there a nightly release already compiled with the HepMC3 flag on?

[andresailer]

No, not yet. Sorry!

[vvolkl]

@eleogran you could try the key4hep nightly build:

source /cvmfs/sw-nightlies.hsf.org/spackages/linux-centos7-broadwell/gcc-8.3.0/key4hep-stack-master-2020-12-07-qca3v5rnrsmeqsrsrb4zkhsi7xattb4k/setup.sh

There dd4hep is already built with HepMC3 ( although I haven't tested the reader myself yet).

[eleogran]

@vvolkl it works!
Thank you and @andresailer