FxFx merging with MadGraph5_aMC@NLO

Introduction:

How to use it with PYTHIA8:

not including QCD light partons nor b quarks (and not including QCD-charged BSM particles)

At the level of the generation of the hard events, make sure to set the following parameters in the run_card:

To include the Sudakov suppression and the assignment of the special renormalisation and factorisation scales in the matrix elements set
3 = ickkw
To not over write the scales, make sure to set the use of fixed scales to false (and use the default SubProcesses/setscales.f)
False = fixed_ren_scale
False = fixed_fac_scale
False = fixed_QES_scale
Set a transverse momentum cut on the jets with the following jet definition
### = ptj
1.0 = jetalgo
1.0 = jetradius
where ### is a number at most half as large as the merging scale.

The merging scale is set by the Qcut parameter
Qcut = ###
And the maximum multiplicity of all the processes included in the merging is set by by the njmax parameter
njmax = ###
where the njmax corresponds the maximum multiplicity of the Born processes included.

As of MadGraph5_aMC@NLO v3.3.0, Pythia8306 should be used, with the pythia8306/include/Pythia8Plugins/JetMatching.h file replaced by this one. (New installations of 3.3.0 will not automatically update the JetMatching.h).

How to use it with HERWIG6:

At the level of the generation of the hard events, make sure to set the parameters related the merging in the run_card in the same way as explained above for PYTHIA8.

Proceed normally with the event generation, but at the level of the showering, events for which the jets after showering do not 'match' the jets before showering should be removed. The matching should be understood in the MLM way. The easiest to do this, is at the level of the analysis. For HERWIG 6 you can find a piece of code

Here

(please make a diff with the default version of mcatnlo_hwan_stdhep.f to see what the changes are)

matching_scale equal to the merging scale
n_match equal to the number of jets at the Born contribution of the current sample
max_multiplicity to true if the current sample is the one with the highest multiplicity considered

Limitations:

Processes with jets or b quarks at the lowest multiplicity cannot be consistently merged.
Use PYTHIA8 version 8.210 or later. (Even though FxFx is available from Pythia8.186, there was a bug for the highest multiplicity sample).
For PYTHIA8, the code does correctly not include any of the decays in the merging procedure when using MadSpin. This is not the default with HERWIG6, but the rejection in the analysis can be updated to also skip these jets.
Only tested with the HERWIG6 and PYTHIA8 parton showers, but should also work for HERWIG++. It does not work for the Q²-ordered PYTHIA6 code.

Reference:

arXiv:1209.6215

Merging meets matching in MC@NLO