MadGraph5_aMC@NLO

Generating events with flat pT distribution

Asked by akash

Dear MadGraph Team,

I am working on machine learning applications for a particular BSM scenario, and as part of my work, I am training different ML algorithms with W-initiated jets obtained from FastJet to build a robust W-tagging algorithm.

The training datasets for these applications come from <email address hidden> generated diboson events (p p > w+ w- > hadronic decay). In order to not bias the training datasets, I would like to generate events such that the pT distribution of the W boson at generator level is a flat distribution rather than the usual one. Is there a way to do this?

Thank you.

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

Should be possible if you know the original pt distribution
(at worse a first run can give it to you).

Then you can use this method:
https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/LOEventGenerationBias

Cheers,

Olivier

Revision history for this message
Eloise Johnston (johnston44) said (last edit ): 		#2

The pT spectrum is a generator thing; eta distribution is - there are a couple of elements that you’ll probably want to tweak in your generator. Or do you have one given particle with one given momentum X, and you want to sample possible decays, with flat eta, etc? Then that sounds like TGenPhaseSpace might help - but IIRC it doesn’t allow you to set boundary conditions like you’re after.

Again - the best idea is probably to talk to someone who knows what you try to do, and advise you how you can get there. With the info, you provide my answer would be: tweak the generator.

When I am entering the field of blogging with the project https://electricbikeslounge.com/, I am facing many difficulties. But I did not lose my energy.

Can you help with this problem?

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

To post a message you must log in.