Using MadGraph in ATLAS

External Software

It looks like ATLAS does not have its own version of MadGraph. Rather, there's an interface that will take output files from MadGraph and put them in the ATLAS event generation pipeline. I've downloaded MG_ME_V4.4.44 and BRIDGEv2.18. The latter is used to calculate decay widths / branching ratios. Don't download the StandAlone version (MG_ME_SA_V4.4.42)!


I tried this without setting up the athena framework.
  • Ran make in MG_ME_V4.4.44/MadGraphII
  • Ran make in MG_ME_V4.4.44/BRIDGE
    • Complained about not finding the HELAS library
  • Ran make in MG_ME_V4.4.44/HELAS
  • Ran make in MG_ME_V4.4.44/BRIDGE
    • Complained about not finding the g2c library
    • Replaced -lg2c in makefile with -lgfortran. SLC5 (new versions of gcc) don't support g2c. OTOH, there is a file /usr/lib64/ Why won't this do it?
    • If I do -lgfortran or -L/usr/lib64/ I get undefined reference to `z_abs' from HELAS/lib//libdhelas3.a(vvvxxx.o): In function `vvvxxx_'. This indicates that it's still not getting the right symbols.
    • Solution is to use gfortran.
  • Changed f77 to gfortran in MG_ME_V4.4.44/HELAS/Makefile
  • Ran make in MG_ME_V4.4.44/HELAS
  • Changed -lg2c to -lgfortran in MG_ME_V4.4.44/BRIDGE/source/makefile
  • Ran make in MG_ME_V4.4.44/BRIDGE... success!

Running MadGraph

I start off in MG_ME_V4.4.44/Models/heavyquark (which I copied from usrmod).

Setting up a new model

  • New particles go in particles.dat
  • New interactions go in interactions.dat
  • Variable names for coupling strengths go in VariableName.dat.

Input: particles.dat, interactions.dat, VariableName.dat

Output:, couplings.f,, printout.f, lha_reading.dat, param_card.dat

This sets up a skeleton. The masses and widths of the new particles, along with their coupling strengths, go in param_card.dat. The user must edit couplings.f to translate the tunable parameters into the vertex interaction strengths (make the connection between interactions.dat and VariableName.dat).

Changing the heavy quark masses

  • Change the masses directly in Models/heavyquark/param_card.dat
  • In this model, the couplings go like v/mQ. If mQ gets bigger, the coupling goes down and vice versa. Looks like the coupling goes to 1 if the mass goes to the top mass (174 GeV). Or maybe it's [Higgs VEV]/sqrt(2). Anyway, change this also in param_card.dat.
  • IMPORTANT: Run touch couplings.f && make couplings && ./couplings
  • Now go to BRIDGE and do runBRI.exe, tell it to run in the model directory (after backing up old param_cards)
  • Give it the string qd qd~ qu qu~ xd xd~ xu xu~ yd yd~ to generate all widths and branching fractions. Since qd CC and qu NC are orthogonal, they can exist in the same param_card.dat file. No need to do three-body decay modes.


  • Now I move to MG_ME_V4.4.44/heavyquark (copied from Template).
  • Edit Cards/proc_card.dat
    • Change sm to heavyquark (from Models directory)
    • Specify process to generate as pp>(qd>(w->l-vl~)j)j
    • Set QCD vertices to 0 and QED vertices to 4.
  • Run bin/newprocess to generate Feynman diagrams (sets up an index.html file in the folder!). They look reasonable.
    • May want to include bb~ in proton/jet definition later
  • Copy param_card.dat from the Models directory into the ./Cards directory. Can change masses and widths here.
  • Run bin/generate_events to generate 10000 events and determine a cross section.
    • This step uses run_card.dat to set the collider parameters. I think I'll need to mess with beam energies and PDFs here.
  • The Events directory will now have an *unweighted_events.lhe.gz file. I copy this to my WorkArea in and follow the directions in MadGraphForAtlas

Pathena Commands

  • Event generation:
     pathena --trf " ecmEnergy=7000 runNumber=777780 firstEvent=1 maxEvents=-1 randomSeed=%RNDM:123456 outputEvgenFile=%OUT.EVNT.pool.root inputGeneratorFile=%IN" --inDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.TXT.v1 --outDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.evgen.EVNT.v1 
    • 5000 events in a job will take 10 minutes and use 45 MB of space.
  • Simulation:
     pathena --trf " --mcinput=True inputEvgenFile=%IN outputHitsFile=%OUT.HITS.pool.root maxEvents=50 skipEvents=%SKIPEVENTS randomSeed=%RNDM:8675309 geometryVersion=ATLAS-GEO-10-00-00 physicsList=QGSP_BERT conditionsTag=OFLCOND-SIM-BS7T-00 DBRelease=%DB:ddo.000001.Atlas.Ideal.DBRelease[BLAH] IgnoreConfigError=False" --outDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.simul.HITS.v1 --inDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.evgen.EVNT.v2/ --nEventsPerJob 50 
    • 20 events in a job will take 2 hours and use 10 MB of space (2.5 GB for 5k events)
  • Digitization/Reco:
     pathena --nFilesPerJob 2 --trf " --mcinput=True inputHitsFile=%IN outputRDOFile=%OUT.RDO.pool.root skipEvents=0 geometryVersion=ATLAS-GEO-10-00-00 digiSeedOffset1=%RNDM:2345 digiSeedOffset2=%RNDM:346621 samplingFractionDbTag=QGSP_BERT triggerConfig=DEFAULT conditionsTag=OFLCOND-DR-BS7T-ANom-15 DBRelease=%DB:ddo.000001.Atlas.Ideal.DBRelease[BLAH] IgnoreConfigError=False; inputRDOFile=%OUT.RDO.pool.root outputAODFile=%OUT.AOD.pool.root outputESDFile=%OUT.ESD.pool.root maxEvents=-1 skipEvents=0 geometryVersion=ATLAS-GEO-10-00-00 conditionsTag=OFLCOND-DR-BS7T-ANom-15 DBRelease=%DB:ddo.000001.Atlas.Ideal.DBRelease[BLAH] " --outDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.recon.AOD.v1 --inDS user.jtuggle.mc09.777780.Pythia_MadGraph_qd_enu_300.simul.HITS.v1/ 
    • 20 events in a job will take 30 min and use 27 MB (6.8 GB) for RDO, 16MB (4 GB) for ESD, 6MB (1.5 GB) for AOD (12 GB total)
    • Note: You must output the ESD or the reco script has issues.

-- JoeTuggle - 30-Sep-2010

-- JoeTuggle - 30-Jun-2010

Edit | Attach | Watch | Print version | History: r8 < r7 < r6 < r5 < r4 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r8 - 2010-11-03 - JosephTuggleExCern
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback