aMC@NLO Matching

The aMC@NLO package [Fri02] attempts to automate the MC@NLO matching procedure [Fri02]. MC@NLO interprets the parton shower as NLO subtraction method, and removes unwanted parton-shower contributions by extending the subtraction scheme used to generate NLO fixed-order results. Upon showering, an NLO accurate prediction for inclusive observables is achieved. This makes MC@NLO a convenient NLO+PS matching scheme. A consistent extended subtraction in the NLO fixed-order result makes analytic knowledge of the shower emission probability necessary. Once this is known, interfacing the (parton-shower specific) NLO calculation with the shower is straightforward.

To allow for a fast, automatic generation of shower subtractions that are used in the fixed-order calculation, Pythia allows to generate emissions with a "global" recoil scheme, in which the recoil of an emission is shared among all final state particles. When using aMC@NLO, this global recoil must be switched on. Please see Timelike Showers for details and options.

A minimal set of settings necessary for a consistent treatment of aMC@NLO inputs is

         SpaceShower:pTmaxMatch = 1

         SpaceShower:pTmaxFudge = 1.

         TimeShower:pTmaxMatch = 1

         TimeShower:pTmaxFudge = 1.

         SpaceShower:MEcorrections = off

         TimeShower:MEcorrections = off

         TimeShower:globalRecoil = on

         TimeShower:weightGluonToQuark = 1

and further (process-specific) settings related global recoils.

Some comments are in order. The settings SpaceShower:pTmaxMatch = 1, SpaceShower:pTmaxFudge = 1., TimeShower:pTmaxMatch = 1, TimeShower:pTmaxFudge = 1. are included to ensure that the correct parton shower starting scale (i.e. the scale set when generating the subtractions in MC@NLO) is used within Pythia. Note that the last three options are default in Pythia8, and that the first option differs from the default only if the input state does not contain final state partons.

Matrix element corrections to the parton shower splitting kernels have to be switched off by SpaceShower:MEcorrections = off and TimeShower:MEcorrections = off . This is necessary because the matrix element corrections are not suitable for showers in the global recoil scheme, and because it is not viable to include process-specific shower probabilities in an automatic framewrok like aMC@NLO.

TimeShower:globalRecoil = on is necessary. Formally, it is allowed to switch back to a local recoil treatment beyond the first proposed emission of any of the hard scattering partons in Born-type events. Pythia offers three choices at which stage the global recoil is dropped in favour of a local strategy. It is necessary to supplement the setting TimeShower:globalRecoil = on by additional settings specifying which global recoil strategy should be used. As these choices are up to the user, please consult Timelike Showers for details on these options.

Finally, TimeShower:weightGluonToQuark = 1 is not default any longer, but was it at the time the subtractions were first implemented, and so is required for consistency until further notice.