Big discrepancy in electron efficiency between Delphes3 and PGS

[Jeff Asaf Dror]

Hi,

I'm using Delphes3 to study the detector effects and I'm finding a huge discrepancy between Delphes3 and PGS in their efficiency to tag electrons. In both cases I'm using the CMS cards and outputting my .hep file using Madgraph on the test process, u u~ > e+ e-, with the default cuts on the run_card (pT >10, abs(eta) <2.5). I attached plots showing the output for number of electrons in each case. but typically I find an efficiency of about 70% for Delphes3 and 90% for PGS.

[Jeff Asaf Dror]

Number of electrons

[Alexandre Mertens]

Hello,

The efficiency for Electron (pT > 10 and abs(eta) < 2.5) is set at 95 % in the default CMS card. Nevertheless, there is an extra isolation criteria that might causes the difference.

How did you compute the "70%" you obtained?

Alexandre

[Jeff Asaf Dror]

Thanks for your response! I found that out of 1000 events, Delphes3 kept about 5000 events with 2 electrons (see plot). This corresponds to an efficiency for tagging an electron of, x*x = 1/2, or x ~ 70%.

Since I am looking at a simple 2 electron event I find it surprising that so many would be rejected by isolation cuts?

[Alexandre Mertens]

I just modified my first comment with the correct values.

It is possible to remove/modify the isolation. To remove it, you can replace the isolated electron input to the UniqueObjectFinder by simple electrons in delphes CMS card.

module UniqueObjectFinder UniqueObjectFinder {
# earlier arrays take precedence over later ones
# add InputArray InputArray OutputArray
  add InputArray PhotonIsolation/photons photons
#  add InputArray ElectronIsolation/electrons electrons
  add InputArray ElectronEfficiency/electrons
  add InputArray MuonIsolation/muons muons
  add InputArray JetEnergyScale/jets jets
}

I hope this can help.

[Jeff Asaf Dror]

I see, thanks. I removed the isolation cut as suggested (I assumed you meant to have an "electrons" at the end of the added line). This did increase the number of electrons but not significantly. Now I'm finding about 15000 electrons in total and given that I started with 20000 (2 per event), corresponding to an efficiency of about 75%. Given the efficiencies you quote above, this still seems too low.