Opened 10 years ago
Closed 10 years ago
#329 closed Bug (fixed)
Low charged lepton efficiencies
Reported by: | Marc Thomas | Owned by: | |
---|---|---|---|
Priority: | minor | Milestone: | |
Component: | Delphes code | Version: | Delphes 3 |
Keywords: | Cc: |
Description
Hi,
I'm running Delphes via MG5_v2.2.1, having installed pythia-pgs and Delphes3 via the "install" command. The installed versions are,
pythia-pgs: PyPGSVersion, 2.4.0
Delphes: VERSION, 3.1.2
When I generate 10,000 Standard Model events of the process
p p > z, z > e+ e-, with PT_lepton > 10. |eta_lepton| < 2.5,
and analyse in Delphes3 using the /examples/delphes_card_ATLAS.tcl as my delphes_card.dat, only 7797 of the events contain an electron, giving an electron efficiency of 78%.
When I analyse the event with pgs and Delphes 2 I get efficiencies of,
Delphes2 - efficiency = 95%
pgs - efficiency = 88%
This 78% efficiency for such a clean event seems very low, both given the Delphes2/pgs results, and looking at the electron efficiencies in the delphes_card, (of 95% for central hard electrons which most of my electrons are, and 85% for larger rapidity up to 2.5).
As a check, I altered both the "electron tracking efficiency" and the "electron efficiency" to 100% in all regions. This increases the efficiency to 94%, which again seems low given the 100% efficiency in the card.
This looks like a bug to me, or am I missing something. Are the isolation cuts in the default ATLAS card too harsh?
(For p p > z > mu+ mu- I also get a low efficiency of 83%.)
Any advice would be welcome.
Cheers,
Marc
Change History (3)
comment:2 by , 10 years ago
Thanks Michele,
It was the combination of electron efficiencies and the tight isolation cuts that was leading to the low efficiency.
Cheers,
Marc
comment:3 by , 10 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
(edited)
Hi Marc,
sorry, forget my last comment, I misread you.
This might indeed be related to isolation cuts being too tight.
If if you want to relax them, you can increase the
PTRatioMax value.
Michele