mapyde 0.1.0

A great package.

mario-mapyde

Docker Images

Docker images are made available in our container registry.

docker pull ghcr.io/scipp-atlas/mario-mapyde/madgraph
docker pull ghcr.io/scipp-atlas/mario-mapyde/delphes
docker pull ghcr.io/scipp-atlas/mario-mapyde/pyplotting

If you want to run on a machine with an NVidia GPU and use it for limit setting with pyhf, then there's a container for that too:

docker pull ghcr.io/scipp-atlas/mario-mapyde/pyplotting-cuda

Running

There are a few layers of scripts to factorize the different tasks. A typical pipeline will look like:

1. Call test/wrapper_mgpy.sh to run MadGraph+Pythia and produce a .hepmc file. The script takes options to specify things like:
   • proc/param/run cards for MadGraph
     • includes specifying particle masses, and
   • any kinematic cuts to apply at parton level
   • pythia card
   • center of mass energy
   • number of cores to use for MadGraph and Pythia
2. Call test/wrapper_delphes.sh to run Delphes, which is a parameterized detector simulation. The output is a ROOT file. The script takes arguments to specify things like:
   • Delphes card
3. Call something like test/wrapper_ana.sh to analyze the Delphes output. Note that this script can run user-specified code:
   • scripts/SimpleAna.py will make a generic "flat" ntuple in a ROOT file.
   • scripts/Delphes2SA.py will make an ntuple that can be parsed by SimpleAnalysis for limit setting.
4. If you want to run limits, then there are two additional steps:
   1. Run test/wrapper_SimpleAnalysis.sh to analyze the output of Delphes2SA.py and make inputs for limit setting
   2. Run test/wrapper_pyhf.sh to plug the results from SimpleAnalysis into the public likelihood.

Each job gets its own ${tag}, which is used to tell the various steps in the pipeline which data to operate on.

For an example of a full pipeline, see run_VBFSUSY_standalone, which itself takes various options to help steer the work.