Madgraph setup for generating Phi(tau tau). Instructions:
- Generate cards with make_cards.py or make_cards_1j.py. The 1j gridpacks are much faster to create, so these are better for debugging.
- Checkout genproductions (in 2021, master uses MG 2.6.5).
git clone git@github.com:cms-sw/genproductions
cd genproductions/bin/MadGraph5_aMCatNLO
- Try creating a gridpack:
- Interactively: genproductions doesn't have an out-of-the-box script, unfortunately, but you can easily make one. Copy the condor script to a new script, e.g.
cp submit_condor_gridpack_generation.sh submit_interactive_gridpack_generation.sh
, and change "condor" to "interactive" in the script. Alternatively, you can run the last line of this script directly. - Using HTCondor:
source submit_condor_gridpack_generation.sh Spin0ToTauTau_1j_scalar_g1_HT300_M100 ../../../cards/scalar_m100/
(args 3 and 4 are SCRAM_ARCH and the CMSSW version, which should be left blank to use the defaults).
- Interactively: genproductions doesn't have an out-of-the-box script, unfortunately, but you can easily make one. Copy the condor script to a new script, e.g.
Notes:
- CMS stores UFO files at
/afs/cern.ch/cms/generators/www/DMsimp*
- Phi(bb) used DMsimp_s_spin0_4f. This appears to be an outdated alpha version... it doesn't work with MG >= 2.6.1, and it doesn't work with HTCondor. It's probably better to use DMsimp_s_spin0.
- We should copy the hacked UFO file with tau couplings here, so the genproductions scripts can find it automatically. You can also modify the scripts to find the UFO manually (look for the
wget
command where the UFO is downloaded), but this is bad for future reproducibility.
- The recommended version of Madgraph is 2.6.5. Newer versions haven't been validated by GEN yet.