Right, so the cutflow table is being used for two things:
1) compare the common framework with an independent framework
2) compare the final numbers from those two with the ntuples (and make sure everyone is using the ntuples in the same way)
So since the ntuples have some preselection, it's not surprising for the ntuple numbers to differ from the non-preselected numbers from the framework. But after cleaning cuts, the MET requirement, and >= 2 baseline leptons, the ntuple numbers should agree well with the framework numbers.
passEventCleaning is here, and you can see it includes GRL, bad muon, bad jet, etc.
In your previous mail, you mentioned that we are relying on the inclusive MET trigger and pointed me the twiki page. But you only pointed 2016 triggers. Because the data should include 2015 and 2016, why don't we apply 2015 trigger
The links were correct, but I had a typo. It should have said here (2016) and here (2015).
And these triggers are applied for the data, how about the MC? When I worked on the SS/3L study, Ximo told me to apply the triggers on the MC depending on the pileup random runnumber. If the pileup random runnumber < 290000, then I have to use 2015 trigger. Otherwise I have to use 2016 trigger. Is there similar step in the Higgsino study?
Ah, good point! You should actually use the branch trigMatch_metTrig, which takes care of this automatically for us. The ntuple only fills the trigMatch_metTrig branch for the appropriate triggers and runnumbers (see here, which relies on this function, though it starts to become a bit detailed). I think I'll copy some of this information over to Jira, so everyone is on the same page.
I also want to ask that at which stage in the cutlfow the trigger should be applied? Right after Bad jet requirement?
I think in the cutflow, we only apply the trigger as one of our non-sequential cuts (which means non-sequential AFTER event cleaning). But maybe it would be useful to add more cutflow entries for nElectrons, nMuons, nJets, etc. at the post-ntuple preselection stage, i.e. after >=2 baseline leptons and >150 GeV MET.
Hi Yu-Ting,
Right, so the cutflow table is being used for two things:
1) compare the common framework with an independent framework 2) compare the final numbers from those two with the ntuples (and make sure everyone is using the ntuples in the same way)
So since the ntuples have some preselection, it's not surprising for the ntuple numbers to differ from the non-preselected numbers from the framework. But after cleaning cuts, the MET requirement, and >= 2 baseline leptons, the ntuple numbers should agree well with the framework numbers.
passEventCleaning is here, and you can see it includes GRL, bad muon, bad jet, etc.
The links were correct, but I had a typo. It should have said here (2016) and here (2015).
Ah, good point! You should actually use the branch trigMatch_metTrig, which takes care of this automatically for us. The ntuple only fills the trigMatch_metTrig branch for the appropriate triggers and runnumbers (see here, which relies on this function, though it starts to become a bit detailed). I think I'll copy some of this information over to Jira, so everyone is on the same page.
I think in the cutflow, we only apply the trigger as one of our non-sequential cuts (which means non-sequential AFTER event cleaning). But maybe it would be useful to add more cutflow entries for nElectrons, nMuons, nJets, etc. at the post-ntuple preselection stage, i.e. after >=2 baseline leptons and >150 GeV MET.
Joey