few events in certain e+e- radiative return reactions

Asked by nblinov on 2020-07-20


I'm trying to generate e+e- > chi chi via a vector mediator (that can go on shell) using the ISR plugin (https://github.com/qliphy/MGISR). Chi is a complex scalar or Dirac fermion.
I run into a strange issue where MG5 fails to generate the requested number of events only for the scalar case but not the
fermion model (about a third of the requested events is generated and extremely slowly)
I checked the standard things: there are no explicit cuts on the final state; the couplings are chosen such that the width of the s-channel vector is not tiny (about 1e-3 - 1e-4 of the mass). The cross-sections estimated by MG5 are equal, as expected for a process with an on-shell intermediate state.
Does anyone know how this issue can arise? Thank you!

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :

This is likely a phase-space integration issue.
I do not know how the MGISR is doing the phase-space integration but this can be very tricky if the mass of the mediator is close to the beam energy. In that context I would not be surprised to have the phase-space integration to fail (but if they put a lot of effort to cure such issues). The best is likely to ask the authors of the plugin.



Revision history for this message
nblinov (nblinov) said :

Hi Olivier,

Thanks for your response. I looked at the plugin code and it does not modify the phase space sampling directly, but only determines new e+/e- momenta using a modified get_dummy_x1_x2 in genps.f
In particular it still looks like I'm still using sample_get_x which has some contingencies built in for BW and s channel resonances.
Naively this seems like an even simpler version of qqbar -> Z, which MG has no trouble with...

On a related note, I noticed that in genps.f there's special handling of 2->1 processes in subroutine gen_mom. Is this specifically for processes that are 2->1 at the level of the proc_card or would the resonance contribution 2->1->2 process count as 2->1?

Thanks for your help!

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :


2>1 and 2>1>2 are quite different.
2>1 is very particular since the 3 particle are onshell, if you do not have PDF the process is actually impossible. With PDF the degrees of freedom is still one dimensional (the longitudinal boost) (3*number_of_final_particle -4 [conservation of energy] + 2 [PDF] = 1)

2>1>2 has more degrees of freedom (3*number_of_final_particle -4 [conservation of energy] + 2 [PDF] = 4).
including the invariant mass of the "1". In the context of lepton collider I expect that a huge contribution is coming from the offshell part of that particle --where the beam are at nominal energy--)



Can you help with this problem?

Provide an answer of your own, or ask nblinov for more information if necessary.

To post a message you must log in.