-- AmnonHarel - 08 Dec 2008

Jet ID Links

Documentation	from JetAlgo
1st official jet ID analysis note page	official jet ID project page
Latest (Forward09) jet ID - the AN	JES for 7TeV MC
from DPGs	Software
Official HCAL anomalous event removal page	SWGuidePATSelectors (for Sal's new versions of JetIDSelectionFunctor)
Quality flags from DPGs - HCAL RecHit flags, ECAL RecHitflags (in the 2nd table) & their caveats!	integration with CMSSW tests
	Temporary hack I used to check jet ID usage in PAT

• Selector.h: Added method printActiveCuts, useful for debugging selectors with many versions...
• JetIDSelectionFunctor.h: Version for cvs - added fwd09 cuts, debugged over a few hundred jets
• JetIDHelper.cc: New version for cvs - documented HF-AOD-nHitsIn90 bug and slightly improved a warning message

Efficiency plots

Inefficiencies of "Pure, loose" jet ID

Pure MC truth technique

See An for details - these are gen particles within the jet cone. New feature - no remapping done, x-axis shows the average pT_gen in each bin. Note: Having a few GeV of generated particles doesn't mean it's a real jet - at very low pT_gen the inefficiencies may be inflated by cases where noise dominates. This should be quite common near the jet reconstruction threshold, which is determined from the cell and tower thresholds.

Region	|eta| range	all pTs	excluding low pT
HB	0 - 1.0
BE	1.0 - 1.8
HE	1.8 - 2.6
dEF	2.6 - 3.2
dHFa	3.2 - 3.8		not needed, see plot on the left
HFb	3.8 - inf.		not needed, see plot on the left

Truth matched jets as a function of reconstructed pT

"Truth matched" means that at least 60% of the reconstructed (L2L3 corrected) energy is visible in stable particles (the same ones clustered into genJets) with dR<0.5 relative to the jet axis. To avoid biases, the efficiencies are measured in bins of the generated pT (same one used for the matching above), and then remapped to the reconstructed pT.

Region	|eta| range	all pTs	excluding low pT
HB	0 - 1.0
BE	1.0 - 1.8
HE	1.8 - 2.6
dEF	2.6 - 3.2
dHFa	3.2 - 3.8		not needed, see plot on the left
HFb	3.8 - inf.		not needed, see plot on the left

Tag & Probe technique

Statistics are a problem at low pT and in the HF.

HB	HB-HE transition	HE
|eta|<1.0	1.0<|eta|<1.8	1.8<|eta|<2.6

HE-HF transition	"central" HF	"very forward" HF
2.6<|eta|<3.2	3.2<|eta|<3.8	3.8<|eta|

BTW: the HFb plot is at 50-80 GeV.

Inefficiencies of n90hits>4 cut (2TeV MC)

As usual, I recommend the tag & probe method results and expect less bias in them. But their reach at low pT is insufficient for current running.

region	Truth based inefficiencies	Tag & probe inefficiencies
HB
HE

Inefficiencies due to cut on maximal EMF

This is in addition to the minimal EMF cut, whose inefficiencies are negligible here (and are detailed in the official jet ID page).

The inefficiencies are shown below for 5 detector regions:
HB is |eta|<1.0, BE is 1.0<|eta|<1.75, HE is 1.75<|eta|<2.6, EF is 2.6<|eta|<3.25, HF is 3.25<|eta|.

--+++ Selected plot(s) on efficiencies of requiring at least one track match, December 2009, 2.36TeV MC - HE efficiencies:

tracktype:	any		"goodIterative" and pT>1
matching done at:	vertex	calo. face	vertex	calo. face
tag & probe
no tag & probe	not ready yet

Jet properties plots

Random plots

JES in 2009

	MC (2.36TeV)	Dec.09 data @ 2.36TeV	Dec.09 data @ 900GeV)
JES vs abs(eta)
pT for abs(eta)>4.2

PAS round circa mid 2009

• production cross section for events passing di-jet selection with various calorimeter only vetoes (green=none, black=preapproval presentation, red=0.1*tagpt) in the HF region:
  

• JetID.pdf: Partial draft

Selected plot(s) from development round circa late November 2008

• jc1227v0_8_manEdit.eps: jc1227v0_8_manEdit.eps

• n90 in HB for various pT ranges:
  

• Simulated inefficiencies from f1HPF<0.85:
  

• JEt eta in min bias 900GeV samples:
  

• Jet (corrected) pT in min bias 900GeV samples:
  

• Jet raw pT in min bias 900GeV samples:
  

* jet_com_bias_900.pdf: Slides on biases in jet commissioning PAS