[Rivet] Missing of Matrix Element particles with status 23 cut

Andy Buckley andy.buckley at cern.ch
Wed Mar 7 13:54:09 GMT 2018


Here you are hitting the problem that the processes you are comparing are not actually the same, and you need to find a mapping between them.

I don't know what is most sensible in this case, but choosing the highest-pT photon (or the one with largest kT-distance from the lepton?) might be a reasonable approach. In general you cannot know which one was "first", since both the shower photons and the ME one are perturbative.
Alternatively, or to validate a heuristic for leading-photon identification, you could turn off the QED shower in Pythia!
Andy
Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
Particle Physics Experiment Group, University of Glasgow

On Mar 7 2018, at 1:03 pm, Yichen <yichen.li at cern.ch> wrote:
>
>
> Hi Andy,
> Thanks for the suggestion.
> Yes, I've considered using PromptFinalState.
> But the problem is there will be additional QED added by Pythia8, in addition to the one in the ME, which will be considered as PromptFinalState.
> And I don't want these additional QED photons, since the high order calculation, which will be used for deriving the kfactor, consider only 1 photon radiation.
>
> But I didn't study this effect quantitatively. Maybe it can be neglected or suppressed with cuts. I'll take a look.
> Best,
> Yichen
>
> On 07.03.2018 10:56, Andy Buckley wrote:
> >
> > Hi Yichen,
> > The accepted way to identify hard-process leptons and photons is by working backward from them in the final state, and verifying that they do not have any hadronic ancestors. This is specified in the ATLAS truth-object definition PUB note, for example, and is most conveniently achieved in Rivet by using the PromptFinalState. Regardless of intentions, starting from looking for "ME" particles with particular status codes will be on the verge of unphysicality unless you are very careful.
> > Hope that helps,
> > Andy
> >
> > Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
> > Particle Physics Experiment Group, University of Glasgow
> >
> >
> >
> >
> >
> > On Mar 7 2018, at 9:26 am, Yichen <yichen.li at cern.ch> (mailto:yichen.li at cern.ch) wrote:
> > >
> > >
> > > Dear Andy and Torbjör,
> > > Thanks a lot for your replies!
> > > By looking for these ME particles, we are not trying to do unphysical measurement. Our motivation is the following:
> > > 1) We unfolded our measurement to a particular decay channel of the ttbar (e.g. to e/mu+jets channel). So we need the decay information of the ttbar to remove other channels so that we can compare the MC sample, which includes all non-hadronic decay channels, with the unfolded result.
> > > 2) We are trying to calculate kfactor for our MC sample in a phase space where there is jet multiplicity cut (as well as lepton/photon pt/eta cuts). The higher order calculation is only at ME level. So to get the kfactor, we want to identify the ME leptons/photons in the MC sample and trace their history until they reach their final states and then use these final states to define the phase space.
> > >
> > > Here is the piece of code in Rivet (for analyzing e+jets channel):
> > > ""
> > > bool foundMEel = false;
> > > bool foundMEnuel = false;
> > > bool foundMEph = false;
> > > Particles all = event.allParticles();
> > > foreach (Particle p, all) {
> > > if (p.genParticle()->status() == 23) {
> > > if (fabs(p.pid()) == 11) foundMEel = true;
> > > if (fabs(p.pid()) == 12) foundMEnuel = true;
> > > if (fabs(p.pid()) == 22) foundMEph = true;
> > > }
> > > }
> > > if (!(foundMEel && foundMEnuel && foundMEph)) return;
> > > ""
> > > I put an example in cern lxplus068 /tmp/yili/example/
> > > If you can't use lxplus, I also put a copy in dropbox: https://www.dropbox.com/sh/5tdl2slccr3dc9u/AACHY9hlB6pvdrpcSql_2M_na?dl=0
> > >
> > > In this example code, the evnt/ dir hosts the HEPMC format of the MC sample. Its corresponding LHE can be found with the name events.lhe
> > > From the events.lhe, you can find that this is a pure e+jets channel sample. So with the above code, every event should pass the cut. But only 24% passed.
> > >
> > > Best,
> > > Yichen
> > >
> > > On 06.03.2018 21:09, Andy Buckley wrote:
> > > >
> > > > Hi Yichen,
> > > > One of the major design features of Rivet is to avoid unphysical search criteria like this. Different generators write hard processes with different status codes and different topologies, so there is no portable way to look for their particles. More importantly, there is nothing physically distinct about "the matrix element" from the rest of the perturbative simulation: you are in danger of "measuring" an artefact of the calculation, rather than something physical.
> > > > On your specific question, I don't know where the problem is, since you've not supplied any example code. But if you are using the Rivet FinalState, for example, then the name gives a hint as to why there are no matrix element particles in it! I guess that you are doing something else, though, if you were able to get what you wanted with Pythia6. Rivet doesn't do any special treatment of generator internal status codes.
> > > > Andy
> > > > Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
> > > > Particle Physics Experiment Group, University of Glasgow
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > On Mar 6 2018, at 6:13 pm, Yichen <yichen.li at cern.ch> (mailto:yichen.li at cern.ch) wrote:
> > > > >
> > > > > Dear Pythia8/Rivet experts,
> > > > > I'm using Rivet to analyze a MG5+Pythia8 ttbar+photon sample.
> > > > > And I'm trying to get the matrix element particles using status==23 cut.
> > > > > But it turns out that in many cases, "status==23" can't capture all the
> > > > > matrix element particles!
> > > > > (I did similar study for Pythia6, where all matrix element particles can
> > > > > be identified with status==3)
> > > > >
> > > > > Do you know what could be the source of the problem?
> > > > > I'm not sure if it's due to Pythia8 or Rivet. So I send this email to
> > > > > both mailing lists.
> > > > > And please let me know if you need more information!
> > > > >
> > > > > Thanks a lot!
> > > > > Yichen Li
> > > > > _______________________________________________
> > > > > Rivet mailing list
> > > > > Rivet at projects.hepforge.org (mailto:Rivet at projects.hepforge.org)
> > > > > https://www.hepforge.org/lists/listinfo/rivet
> > > > >
> > > >
> > >
> > >
> >
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20180307/74c1e912/attachment.html>


More information about the Rivet mailing list