This code is intended as a plugin to the High Energy Phsyics code MadGraph5_aMC@NLO (v2.6.6+).
It offers a new output mode for the standalone computation of Matrix Elements of scattering amplitudes. This special MG5aMC_PythonMEs
format is in pure Python.
Copy this project in the PLUGIN
folder located in the root directory of your MG5aMC
distribution (v2.6.6+).
The MG5aMC
example script test_MG5aMC_PythonMEs.mg5
can then simply be run as follows (from within the root directory of MG5aMC
):
./bin/mg5_aMC --mode=MG5aMC_PythonMEs PLUGIN/MG5aMC_PythonMEs/test_MG5aMC_PythonMEs.mg5
The Python/TensorFlow code for this example selection of Matrix Elements will be generated in the folder <MG5aMC_root_dir>/MG5aMC_PythonMEs_output_example
and its usage should be self-explanatory from reading the driver script at <MG5aMC_root_dir>/MG5aMC_PythonMEs_output_example/check_sa.py
.
Over time, the goal is to adapt this plugin so that its output can be directly incorporated in ML engines, such as in a TensorFlow
computational graph.
To this effect the current repository provides the additional format TF
for the output
command.
For now, this output behaves exactly identically to the Python
output format (see by yourself by running test_MG5aMC_TFMEs.mg5
), but uses separate daughter classes and templates (suffixed with TF
) which are ready to be specialised as needed for the output code to be directly inserted in TensorFlow
. In particular the driving script template check_sa_TF.py
should be turned into a demo Jupyter
notebook demonstrating the integration within TensorFlow
.
Note that for now the independent parameters of the model are hard-coded to their default value in the parameters.py
script. Eventually we may want to add a facility for reading an SLHA input card, but this is not needed for now.