Scope of the analysis
LO QCD prediction relies on the production of di-jet events, in principle balanced: the extra jets come from radiation and showering. NLO predictions allow different topologies, in principle with harder extra jets. In this case, the evaluation of two ratios (to be evaluated among spectra):
can be sensible to the difference between LO and NLO predictions (in pure LO, both are 1). Actually, depending on the thresholds used in constructing HT, the ratios can be sensible to the low or the high pT part of the jet spectrum: this can be useful also for comparing different MC prediction, even at LO.
The idea for this work came from
http://arxiv.org/pdf/1006.2144v1
Analysis
Samples
By now, the used samples are the pat-tuples produced for
B bbar correlation studies.
JetMET
skims/PD/SD are used.
Analysis Code
Selections
AK5 PFJets are used, with Spring10 corrections (residuals included). Selections comprehend:
- 1st jet with |eta|<2.6 and pT>56,84,120 GeV for HLT_Jet15U, HLT_Jet30U, HLT_Jet50U
- jets with |eta|<3 and pT>XX , where XX can take values from 10 to 70 GeV
- at least two jets in the event
- JetID loose is applied to all jets
The HT variable is build with the selected jets. Furthermore, different HLT path are combined using Jet1 pT windows:
[56,84), [84,120), [120, +inf)
Systematics
A 10% gaussian smearing is used for evaluating Jet Energy Resolution.
A factor
5% + 2%*|eta|
(summed in quadrature) are used for Jet Energy Scale.
Both are applied on data, and then the difference between the modified and the original distribution is summed in quadrature bin by bin and taken as systematic uncertainty.
Results
Current results are quoted for 1.1/pb, from Runs 132440-143731
Jet pt 20
HT over Jet1+Jet2 pt
HT over 2*Jet1 pt
Other plots (HT, Jet1,2 eta/pt, Jets eta/pt, nJets) are available
here
Comments
Pythia8 discrepancy
- Both HT over 2Jet1Pt and HToverJet1Jet2Pt show a good agreement with data (within uncertainties) all over the spectrum, but
pythia8
, which starts to diverge very soon
- Pythia8 behaviour is confirmed in HT spectrum and in the nJets plots, where an jet excess is clearly visible
- From JetPt spectrum, is visible that Pythia8 creates more low energy jets than the others -> this explains the discrepancies. The Jet1 and Jet2 pT spectra are compatible.
Alpgen pT
- Alpgen seems to produce an harder jet2 spectrum, see R1 vs R2 and jet2 pT spectra
Eta shapes
- inclusive jet eta shapes are well described by all MCs (see Data normalized plots)
- Nevertheless, HergiwJimmy fails to describe the leading jet, producing an innatural plateau in the central region
- For the Jet2 eta shape, the opposite is true, with herwig more central and Pythia with some more plateau: in this case no MC well describes data
- Alpgen well describes all jet1,jet2, incl. jet eta shapes
- This is confirmed in MC (see here) and in an independent cross check here
Jet pt 50
HT over Jet1+Jet2 pt
HT over 2*Jet1 pt
Other plots (HT, Jet1,2 eta/pt, Jets eta/pt, nJets) are available
here
Comments
* Some hint of discrepancy in Ht over Jet1+Jet2 pT, even at low energy (see also the zoomed plot)
* Sill discrepancies with Pythia8
* Same considerations as above for eta shapes
* Still Alpgen harder Jet2 pT spectrum creates bigger differences between R1 and R2
Jet pt 70
HT over Jet1+Jet2 pt
HT over 2*Jet1 pt
Other plots (HT, Jet1,2 eta/pt, Jets eta/pt, nJets) are available
here
Comments
- At high pT there seems to be some discrepancy. More statistics needed (both for stat and syst uncertainty), but seems promising. NEED TO CHECK WITH NLO. Alpgen seems to describe it better
- For both ratios, Pythia8 discrepancy is smaller, but still there. The same holds for HT, nJets
- Pythia6 D6T pt distributions are not really satisfactory: excess in low pt region
Eta shapes
- Inclusive eta spectrum has a double peak (already seen here), and from Jet1 and Jet2 spectra it seems one right (left ) peak comes from Jet1 (Jet2).... TODO: TRY SPRING10_DATAV2
- Still, Herwig cannot describe Jet1 eta, and no MC can describe Jet2.
References
Old page (only for logging reasons)
--
LeonardoSala - 16-Sep-2010