search

cms-sw / genproductions

Generator fragments for MC production
https://twiki.cern.ch/twiki/bin/view/CMS/GitRepositoryForGenProduction
79 stars 788 forks source link

Pythia fragment #3781

Open MohitS704 opened 1 month ago

MohitS704 commented 1 month ago

Updating the PYTHIA filter used for gluon fusion off-shell Higgs using MCFM+JHUGen to use the MCMultiParticleMassFilter, which was introduced here: https://github.com/cms-sw/cmssw/pull/45344.

Since MCFM+JHUGen gridpacks are separated by final state, this formulation of the PYTHIA filters separated by the 4-lepton final-state works for each given gridpack. You can see a presentation on this at an HZZ meeting here: https://indico.cern.ch/event/1433234/#38-mc-simulation-gluon-fusion.

There is no filter planned for tau decays - so electrons and muons will be the only ones with a filter applied. For tau decays, simply use the fragment without a filter (GluGluToZZTo4L_TuneCP5_13p6TeV-MCFM-jhugen7.5.2_nofilter.py).

bbilin commented 1 month ago

Hi,

Thanks, we went through the presentation to understand the details, but unfortunately failed to do so.

Why do you want to turn off MPI? If there is a specific physics need that requires such a manipulation, you are invited to present in a GEN meeting such that we can understand the needs.

Thanks a lot,

B&L

gritsan commented 1 month ago

We did make a presentation at the GEN meeting back in October 2023, and Steve Mrenna suggested this particular configuration to have MCFM+Pythia and Pythia+Pythia in good agreement. We are not doing anything different from what Steve recommended. We assumed that this has been settled since almost a year ago, and now we proceed with the actual simulation of Run3 configuration, after implementing an additional filter.

smrenna commented 1 month ago

Everything has to be understood in context. At the time, we were trying to validate the transverse momentum of the hard process, which is somewhat decoupled from the MPI. Because MPIs slow down the code considerably, I suggested we turn it off for the sake of validation of the hard process, only.

smrenna commented 1 month ago

This is from the email of Sep 7, 2023: "If you set the runtime command PartonLevel:FSR = off, then you will not have to deal with the complication of any FSR. You would also turn off MPI in the same fashion."

But this is not for real physics.

MohitS704 commented 1 month ago

I see - then this is a misunderstanding on our end that I will fix. Thank you for weighing in @smrenna!

MohitS704 commented 3 weeks ago

Pinging @bbilin to look over this again, as now MPI is not turned off.

lviliani commented 3 weeks ago

What about turning off BeamRemnants:primordialKT and SpaceShower:MEcorrections? Is this something that you need or did you turn them off just for initial validation purposes?

Also, 'SpaceShower:alphaSvalue = 0.118' and 'SpaceShower:alphaSorder = 2' are redundant settings, because they are already included in the CP5 tune parameters (here). So I think you can safely drop them,

smrenna commented 3 weeks ago

How can I actually see the fragment?

lviliani commented 3 weeks ago

How can I actually see the fragment?

You mean the fragments added in this PR? There's a "Files changed" tab in the top bar, there you can see all the changes.

MohitS704 commented 3 weeks ago

Turning off primordial kT was for testing purposes only. We did a simplified study to see how to make the Pythia shower give a result that was comparable to Powheg (if I remember correctly), so we turned off certain features that added noise. This should not be turned off for production.

Just to make sure - is this also true for SpaceShower:MEcorrections?

smrenna commented 3 weeks ago

Thank you for the question. I think the MEcorrections needed to be off so that we got the correct behavior. I will double check our email exchanges.