massless QED shower leaves lepton pairs with too little inv. mass to be put on their mass shell

[marek]

If both the QED shower and fragmentation are turned on, splitting of the type P->ll leave the lepton pair with too little inv. mass to be put on their mass shell. In case of l=tau, the fraction of events is significant (~1/3 of all such splittings in the attached setup, id to #81). The issue can be circumvented by setting the tau massive. However, this only goes unpunished as long as there is no tau in the ME. Thus, reseting the tau's default to be massive is undesirable as long as the QED shower is not used. Alternatively, the routine to put all particles not entering the hadronisation, i.e. taus, on-shell needs to use some momentum from the strongly interacting particles, hence, spoiling their perturbative picture to some degree. Any ideas?

Marek

[Stefan Hoeche]

I would suggest to use the option where the hadronisation puts the tau on-shell using momentum from strongly interacting particles, if necessary. The tau is a lepton, but it is also a hadron because of its large branching ratio into hadrons. It can therefore be treated like a massless quark in the parton shower, as long as its mass is irrelevant at the energy scale of the hard process. In this case the kinematics is approximate anyway, making it less questionable to shuffle momenta between the tau and other particles. If we are interested in the precise kinematics, we should set the tau massive from the beginning, such that there is no ambiguity when making the transition from the perturbative to the nonperturbative event phase. This would be the same reasoning as for the treatment of quarks.

[marek]

As Signal_Process_FS_QED_Corrections and, thus, the routines setting all particles on-shell for fragmentation do not have any knowledge of the QCD part of the event implementing the above would be rather messy. Instead a warning is issued detailing the particles that could not be put on-shell and advising to set the particles massive in the perturbative calculation if relevant to the user.

[Stefan Hoeche]

Milestone old deleted