A factorisation-aware matrix element emulator (FAME) for NLO QCD K-factors

This is the project repository to accompany the article {}. In the article we describe in detail the construction of an emulator built using neural networks to emulate electron-positron annihilation into 5 jets for NLO QCD K-factors. The emulator uses the factorisation properties of matrix elements, along with universal singular functions (in this case antenna functions described in the article Antenna Subtraction at NNLO) to achieve high levels of accuracy.

Here we provide Python code to replicate our strategy.

Requirements

FAME-Antenna is mainly written in Python. See setup.cfg for more details on packages required.

In the example notebook, we use LHAPDF for $\alpha_{s}$ evaluation and MadGraph for matrix element evaluations. These can of course be substituted with other packages.

Installation

1. Clone the repo
2. Install with pip:

   pip install -e .
   

Usage

An example Jupyter notebook is provided in

src/fame_antenna/notebooks/antenna_model_notebook.ipynb

which runs through the necessary steps to construct the emulator.

License

Distributed under the GPLv3 License.