search

ytatus94 / Higgsino

The NUHM2 Higgsino study
0 stars 0 forks source link

Joey's mail 20170526 #11

Open ytatus94 opened 7 years ago

ytatus94 commented 7 years ago

Hi Yu-Ting,

Quick answer since I'm on my phone: the genWeight has the xsec built into it, so you are double counting the xsec.

I'll look closer in a bit to make sure everything else makes sense though!

Joey

ytatus94 commented 7 years ago

More information: our preselection for the ntuples is just two baseline leptons + a MET > 150 GeV cut; see here (the lines before this are just for cutflow purposes). So you do need to do signal lepton selection.

In case it's useful for comparisons, an example cut string for a SS CR is here (and you can easily modify it for other regions):

"(met_Et>200 && jetPt[0]>100 && (HLT_xe70 || HLT_xe90_mht_L1XE50 || HLT_xe100_mht_L1XE50 || HLT_xe110_mht_L1XE50) && nJet30 >= 1 && nJet30 <= 4 && nLep_signal == 2 && nLep_base == 2 && lep1Charge == lep2Charge && DPhiJ1Met > 2.0 && met_Et/(lep1Pt+lep2Pt) > 5.0 && (MTauTau < 0 || MTauTau > 160) && nBJet30_MV2c10 == 0 && lep1Flavor == 1 && lep2Flavor == 1 && met_Et > 0) * pileupWeight * leptonWeight * eventWeight * genWeight * bTagWeight * jvtWeight * 36075"

This is taken from here, with a few small modifications.

Joey

ytatus94 commented 7 years ago

Hi Yu-Ting,

I don't know if there is a Jira about this, but you could open one. It might be useful for people! By the way, a cutflow challenge exists at HIGGSINO-19, which might be worthwhile to fill out.

But I don’t know what are the signal lepton requirements. I guess the requirements should include same flavour, opposite sign and number >=2 for both baseline and signal but I might wrong.

For the signal region, yes, a SFOS pair is used, with (currently) exactly two baseline and two signal leptons. With those requirements, you can ignore lep3 and lep4. There are other regions where a same-sign or different-flavor pair might be desirable.

We're relying on the inclusive MET trigger, which you can see on the lowest unprescaled page here (2016) and here (2016). Looking at this, I think HLT_xe100_mht_L1XE50 can actually be removed from the cut string.

And I notice that the cut string uses met_Et/(lep1Pt+lep2Pt) instead of using METOverHT. Does this mean I shouldn’t use METOverHT but use met_Et(lep1Pt+lep2Pt)?

Ah right, we've been a bit loose in describing this variable. In the ntuples, METOverHT uses HT defined as the scalar sum of jet pT. But the variable we want to use is the MET over the scalar sum of lepton pT. This is actually stored in the ntuple as METOverHTLep12 (I think this variable was only added in for the v1.4 ntuples, and I copied the cut string from Jesse while he was still using v1.3).

This cut string only uses lep1 and lep2 but I didn’t see lep1Signal ==1 and lep2Signal ==1 requirements. Is it because of CR? Do I have to apply the lep1Signal and lep2Signal for SR and how about lep3 and lep4?

Good eye, lep1Signal and lep2Signal should also be applied.

Joey