JHUGen MELA  JHUGen v7.5.6, MELA v2.4.2
Matrix element calculations as used in JHUGen.
madMELA

Our approach to matrix elements in Madgraph is different to that of EFT2Obs. In keeping with the way that MELA is structured, there are a select few processes/production modes that are supported through the native MELA ecosystem. All the final states are 4-leptons. This is colloquially referenced as "MadMELA" throughout the rest of the text, and is managed by MadMela.hh.

So far, these production modes (and their respective processes) are:

Production Process
quark-antiquark production Signal
quark-antiquark production Background
quark-antiquark production BSI
gluon-fusion Signal

Where

  • "Signal" indicates the presence of a Higgs Boson,
  • "Background" a process with the same initial and final state but without the presence of a Higgs
  • "BSI" the sum of signal, background, and interference between the two

Configuration

MadMELA is configured through a series of Wilson Coefficients and functions. These coefficients are set in the same manner as the couplings used in JHUGen or MCFM, but are all real-valued.

C++

In C++, couplings are set using the SelfDSMEFTSimCoupl array (Mela::SelfDSMEFTSimCoupl), with indices corresponding to the indices shown here.

Python

In Python, couplings can be set using SelfDSMEFTSimCoupl, similar to C++. However, Python utilizes named couplings to self-configure. These can also be seem in the same table referenced above.

The documentation for setting various items in Python can be seen in here

Compilation

MadMELA is compiled through combining multiple different Madgraph stand-alone reweighting areas into one shareable library that is used by the C++ Mela interface through MadMela.hh.

The steps are as follows:

  1. Each different sub-area is compiled to form a static library for that process.
  2. While each area is being compiled, a large common block that contains all the possible parameters for each area is created. This large common block contains all the numbered general couplings that are possible for each Madgraph reweighting area.
  3. This large common block replaces the respective common blocks for each area through symbolic links pointing to the large block placed in a separate dedicated directory.
  4. All the compute functions and other commonly-named items for each given area are renamed in the symbol table using objcopy to something else unique and appropriate.
  5. All of these are collated together and compiled