MadGraph5_aMC@NLO

Longitudinal WZ scattering: failed to generate enough events

Asked by Hannes

Hi,

we are trying to generate electroweak production of WZjj (vector boson scattering) with longitudinal W and Z bosons but cannot generate the requested number of events.

generate p p > w+{0} z{0} j j QCD=0
only generates about 2000-3000 when 10k are requested

generate p p > w+{0} z{0} j j QCD=0, w+ > l+ vl, z > l+ l-
only gives us a few hundred.

We are using the LO SM with no b-mass, default MG cuts, and have tried to change sde_strategy and increase hard_survey to 2 but this has not improved the situation. We tested in versions 2.9.9 as well as 3.5.2.

Is there a way to generate more events for this process, are we missing something?

Cheers,
Hannes

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Did you set a cut on MMJJ to at least 100GeV (to remove the ill defined WWZ and ZZW production.
With such cut, they are no issue for such process:

Cheers,

Olivier

> On 17 Apr 2024, at 16:20, Hannes <email address hidden> wrote:
>
> generate p p > w+{0} z{0} j j QCD=0

Revision history for this message
Hannes (hannes3) said : 		#2

Hi,

this (adding mmjj>200) indeed helped for the variant without decay, with the decay chain we still have the same problem.

Cheers,
Hannes

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#3

Hi,

Huum this is weird, in 4 flavor computation everything is fine.
But in five flavor computation the phase-space has trouble to integrate
q b > q b W+ z (+decay)

I thought that it was related to the top mass, but removing diagram with the top does not help and the problematic channel of integration are not top related...

additionally this syntax was working to generate 10k on its own:
generate p b > w+{0} z{0} j b QCD=0, w+ > l+ vl, z > l+ l-

So I guess that the issue is that the importance sampling between the various channel is too complex too handle.

I'm checking various option now to test this and to check what would be a viable option.

Cheers,

Olivier

> On 18 Apr 2024, at 00:25, Hannes <email address hidden> wrote:
>
> Question #809226 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/809226
>
> Status: Answered => Open
>
> Hannes is still having a problem:
> Hi,
>
> this (adding mmjj>200) indeed helped for the variant without decay, with
> the decay chain we still have the same problem.
>
> Cheers,
> Hannes
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#4

One solution that it is working is to use
set group_subprocesses False

before the generation of the process,
this makes the directory quite large and the survey super slow but it does reach target in that mode

Revision history for this message
Hannes (hannes3) said : 		#5

Thanks Olivier, we will check if this works for us.

Can you help with this problem?

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

To post a message you must log in.