Zero weights in ~1/2 of sample

Asked by Jiri Kvita on 2019-06-03

Dear experts,

I am trying to generate matched Z' and Z'+jets with MG5_aMC_v2_6_4 with the following commands

import model Abelian_Higgs_Model_UFO
generate p p > zp > t t~, t > j j b, t~ > j j b~ @0
add process p p > zp > t t~ j, t > j j b, t~ > j j b~ @1

and at the Delphes (3.4.1) level I see ~1/2 of Event_weight are zero, rest are nonzero, in summery for the 10k test samle e.g.

   4113 evts: w=0
   5873 evts: w=4.03607e-08

Is this expected? Should I really ditch events with zero weights? I do not know how to trace the problem down to whether this is a mg5 or Delphes related question.

Cheers and many thanks,

Jiri Kvita

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2019-06-03
Last query:
2019-06-03
Last reply:
2019-06-03

Yes this is normal, the weights are set to 0 due to the matching/merging procedure (in this case MLM).
The advantage of setting the weights to 0 is that you can change the merging scale without the need to rerun the parton-shower.
Quite useful when you do not know what merging scale to use.

Cheers,

Olivier

Jiri Kvita (qitek) said : #2

Thanks, Olivier! So the zero weights reflect the physics of the matching and double counting and should be respected, then? Just to be sure;) Thanks a lot! Jiri

Correct.

Olivier

> On 3 Jun 2019, at 15:27, Jiri Kvita <email address hidden> wrote:
>
> Question #681210 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/681210
>
> Status: Answered => Open
>
> Jiri Kvita is still having a problem:
> Thanks, Olivier! So the zero weights reflect the physics of the matching
> and double counting and should be respected, then? Just to be sure;)
> Thanks a lot! Jiri
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Jiri Kvita (qitek) said : #4

Thanks Olivier Mattelaer, that solved my question.