1. Dijet Azimuthal Distributions at DØ
Amnon Harel
University of Rochester
for the DØ Collaboration.
Joint Meeting of Pacific Region Particle Physics Communities QCD Session Honolulu, October 29th – November 3rd 2006

2. The Observable Lowest order in pQCD: back-to-back dijets
Jets ordered by pT
Lowest order in pQCD: back-to-back dijets
Jets with equal pT and ΔΦ=π
Jet 1
Jet 2 x y
Jet 1
Jet 2 x y
Jet 1
Jet 2 x y
Additional soft radiation can cause small azimuthal decorrelations
Different pTs; ΔΦ slightly less than π
Additional hard radiation can lead to an additional observable jet and to more decorrelation
ΔΦ<π
Tests O(αs)4 calculations
=NLO
=LO
: Three jet production
: Four jet production
It is not necessary to reconstruct the additional jets in order to probe higher orders!
Amnon Harel

3. The Jets
High energy collisions result in collimated particles, which are clustered into jets with various jet algorithms.
Used the Run II Midpoint Cone Algorithm with ΔR=0.7 at all levels.
seed based
midpoint seeds give infrared safety.
iterative
The calorimeter jet energies are corrected to the particle level by scaling and unsmearing. CH
Calorimeter jet
hadrons FH γ EM
“particle jet”
Time
“parton jet”
Amnon Harel

4. Azimuthal correlations stronger at high pT
The Data
Azimuthal correlations stronger at high pT
Integrated Luminosity ~150pb-1 (first 14 months of RunIIa, )
The analysis is not statistics dominated.
Four leading jet pT bins were used, each corresponds to a trigger that is at least 99% efficient there.
The 2nd leading jet pT>40GeV.
Both leading jets are central |y|<0.5
Quality cuts:
require a central, high quality primary vertex
veto high MET (cosmics)
good running conditions
jet ID to reduce electronics noise, etc.
PRL 94, (2005)
Amnon Harel

5. Selected Experimental Issues
Measuring a ratio reduces uncertainties (also theoretical ones).
Correcting the result to the particle level:
Unfolded using MC (Pythia) with highly parametrized smearing
Angular resolution (better than 20mrad)
pT resolution effects are indirect but bigger, especially at ΔΦ<2.8 where they can change which jet is the 2nd leading one.
Similarly, evaluated the uncertainties due to residual dependencies (in η and Φ) in the jet energy scale that may lead to choosing a different 2nd leading jet.
Additional pp interactions per bunch crossing are negligible.
calculated from cross sections and verified with #P.V.s
Fake jets are most apparent at ΔΦ=~½π, verified that they were all removed.
Amnon Harel

6. Non-Pertubative Effects
 Non pertubative effects are less than 5%
This simplifies comparisons with pQCD calculations, and makes this measurement extremely useful for tuning pQCD in Monte Carlo simulators.
Amnon Harel

7. Data vs. Calculations Calculated with αs(Mz)=0..118
αs3: Diverges due to missing soft processes
αs3: Only three partons
αs4: Resummation Needed
Calculated with αs(Mz)=0..118
αs4: Is only LO here  Larger S.F. uncertainties
αs4: Is only LO here
PRL 94, (2005)
Amnon Harel

8. Data vs. Pythia and Herwig
22 Hard Processes, 3rd and 4th jets are produced by parton showers.
HERWIG describes the data well.
PYTHIA:
Default: not enough jets with low ΔΦ.
Pythia provides many tuning handles, which ones should we use?
Insensitive to non-pertubative effects.
More ISR can help…
Adjusted ΛQCD so that αs(Mz)= used 2-loop solution of normalization group equation (even in Pythia)
PRL 94, (2005)
Amnon Harel

9. Agreement still not perfect at high ΔΦ
Tuning Pythia
Parameters that increase Initial State Radiation
Parameters that increase Final State Radiation
Agreement still not perfect at high ΔΦ
Adjusted ΛQCD so that αs(Mz)= used 2-loop solution of normalization group equation (even in Pythia)
Spell out ISR/FSR?
Increased virtuality
PARP(67)=12.5
Zoomed in to right quarter
Zoomed in to right half
Amnon Harel

10. Matrix element generator Parton shower generator
Matched MC
Good at generating hard, large-angle processes (calculates interference)
LO calculations for 2N hard processes
Matrix element generator
Alpgen / Sherpa
Partons are matched to parton-shower jets to avoid double counting of equivalent phase space configurations.
Weak on the “texture” of the QCD radiation
Hard scatter partons
Good at generating the details within a jet
Only 2->2 hard processes
Parton shower generator
Pythia / Herwig
Multijet events don’t describe data well (and are hard to generate)
Resummed soft, collinear radiation.
Adjusted ΛQCD so that αs(Mz)= used 2-loop solution of normalization group equation (even in Pythia)
Particles
Detector simulation
Dijet ΔΦ is a great testing ground for MC matching:
Only jets
Explore many-jet final states while reconstructing only two of them.
Amnon Harel

11. MC Matching
Alpgen uses the MLM matching scheme. Can use either Pythia or Herwig for the parton showers. Sherpa uses CKKW matching and its own parton shower mechanism.
Phase space is partitioned by number of partons.
The partitions are arbitrary, not physical! They need to be added up (with the right weights) to recover any physical prediction.
Reference: Chapter 4.1 of TeV4LHC QCD Group Report, hep-ph/ , by Begel, Wobisch & Zielinski
Amnon Harel

12. Data vs. Matched MC
Amnon Harel

13. More Work on MC Tunes
Rick Field produced a new global Pythia tune, the “DW” tune, that incorporates the ISR tuning suggested by this measurement.
See Chapter 4.2 of TeV4LHC QCD Group Report, hep-ph/ , by R. Field.
It was shown that the measurement is insensitive to the multiparton interaction model
In Pythia the model was changed (MSTP(82)=4).
Added multiparton interactions to HERWIG using JIMMY.
See “Preparing for measurements of dijet azimuthal decorrelations at ATLAS”, ATL-PHYS-PUB , by A. Moraes et. al.
Adjusted ΛQCD so that αs(Mz)= used 2-loop solution of normalization group equation (even in Pythia)
Amnon Harel

14. Future DØ Plans Redoing the analyses with:
Increased data set: ~7 times the luminosity
Vastly improved jet energy scale:
systematic uncertainties reduced by a factor of ~5!
better coverage of forward regions
Better understanding of the detector may improve other systematics
May add the forward regions: sensitive to BFKL predictions.
Though this measurement requires little statistics and is fairly insensitive to uncertainties on the jet energy scale, dramatic improvements are possible for both. As they were the two leading experimental uncertainties, we expect to achieve a significantly more precise measurement.
Rephrase: Not very sensitive to lumi & JES, but these improvements are so big that….
Amnon Harel

15. ΔΦ at the LHC The tuned MC should prove useful at the LHC
This is a good “week 2” measurement.
Can make a useful measurement:
even with a partially calibrated detector (e.g. at cell level)
even with a very rough jet energy scale that is only for central jets
even with little statistics
Use of leading jets prevents problems with multiple interactions.
ATLAS is preparing to do it.
Amnon Harel

16. Back up slides
Amnon Harel

17. Unsmearing Corrections
The correction is the ratio of the number of MC events (per bin) before smearing to the number after smearing
pT smearing dominates below ~2.8
ΔΦ smearing dominates near π
Amnon Harel

18. The DØ Detector
Amnon Harel

19. MC vs. Calculation
Amnon Harel