Dؒs Recent Results and Experiences with kT and Cone Jet Algorithms

1
DØs Recent Results and Experiences with kT and
Cone Jet Algorithms

• John Krane
• Iowa State University

• Jet Algorithms
• cone
• kT
• kT Inclusive Jet s
• Hadronization effects
• Cone drift
• kT Dijet Thrust

Calor 2002, Pasadena CA
2
A timeline

• UA2 fixed size cones, R1.3 30 cross
  section uncertainty
• TeVatron Run 1 cones R0.7, merge/splitfactor
  of 2 or more improvement in precision
• End of Run 1 DØ looks at kT algorithm, similar
  to ones used by HERA experiments
• CDF/DØ attempt to improve both algorithms and
  achieve consistency

3
The DØ Calorimeter
y

• Liquid argon active medium and uranium absorber
• Hermetic with full coverage
• h lt 4.2 l int gt 7.2 (total)
• Transverse segmentation
• (towers) Dh x D? 0.1 x 0.1
• sE / E 15 /ÖE for electrons
• sE / E 45 /ÖE for pions

q
j
Z
x
4
Jet Algorithms

• Fixed Cone (RunI)
• Iterative
• Fixed cone of radius R
• Overlapping cones split/merge parameter
• Sensitivity to soft radiation

• kT (Ellis-Soper)
• Recombinant
• Distance parameter D
• Infrared and collinear safe in principle

5
kT Algorithm at DØ (Run I)
(Ellis-Soper PRD 48 3160)
A cone jet is just the highest-ET stable cone
6
Cone Algorithm Details (Run I)

• Draw a cone around a seed
• Calc sum ET, and ET-weighted position
• Draw new cone here and recalculate sum ET,
  position
• Reiterate until stable

7
kT Algorithm Details (Run I)
For each particle or pair of particles, starting
at low pT
Is less than ?
Yes
Merge ij
No
Move i to list of jets
Anyleft?
Yes
No
8
Energy Scale
calorimeter jet

• Correction back to the particle level
• Remove noise, underlying event,extra pp
  interactions
• Correct for detector response
• Undo misassignment of particle energies to jets

p
K
particle jet
Time
parton jet
9
Run I DØ Analyses using the kT algorithm
Subjet Multiplicity of Gluon and Quark Jets Phys.
Rev. D65, 052008 (2002) Inclusive Jet Cross
Section Phys. Lett. B525, 211 (2002) Dijet
Transverse Thrust Cross Sections paper in
preparation
KT uniquelywell-suited
Comparison tocone results
Most recent,IR safety important
10
Inclusive Jet Cross Section

• First analysis you docount jets in pT bins
• Central region has large cross section,
  well-controlled systematic uncertainties

11
Results for kT and cone
Lum error removed from plot

• Each distribution iscompared to itsown
  prediction
• Uncertainties highly-correlated fromone bin to
  the next normalization not well-determined, but
  shape is
• Important deviation from cone and from
  predictions at low-pT

12
What causes this difference at low pT?
DØ Data

• Match kT and cone jets in space
• kT jets include more energy than cone
• Insert this difference into the cross section and
  they match!
• So why the energy diff?!

DØ Data
13
Energy diff. in MC

• MC energy difference is half the size, accounts
  for half the s shift (just tuning?)
• Why does kT gather more energy than cone for
  particles, yet match at parton-level?
• particle includes hadronization and spectator
  particles
• Cone gives up quite a bit of E during
  hadronization
• kT seeks and finds had. products and even more

14
What shall we do with this kT result?

• Underlying event estimated with Monte Carlo plus
  minbias overlay would a beampipe jet do a
  better job?
• Theory (PDFNLO) grew up with conethey match by
  design PDFs reflect cone algorithm not
  necessarily physics?

15
Lost Energy (cones slip away during iteration)

• Currently a CDF study
• DØ is having trouble isolating examples of this

E
Two jets
Merged jet
Lost jet
16
Suggested improvement to the cone algorithm
(New algo must be valid for theory too!)

• New way
• Draw smaller cone r around seed
• Sum energy in cone,find ET-weighted centroid
• Repeat (2) unless centroid is same as last
• Increase cone size to R, sum energy again but
  keep current centroid
• Place cone of size R at midpoints of the seeds,
  sum energy, find centroid
• Repeat (5) until cone is stable
• Discard duplicates

• Old way
• Draw cone R around seed
• Sum energy in cone,find ET-weighted centroid
• Repeat (2) unless centroid is same as last
• Place cone at midpoints of surviving jets, sum
  energy, find centroid
• Repeat (4) until cone is stable
• Discard duplicates

17
Dijet Transverse Thrust

• Sum over leading 2 jets only
• Split data in several samples based on HT3
• Minimal sensitivity to noise (but beware angular
  resolution)

T1 (always)
2 partons in final state
T2/3,1 (Jetrad)
3 partons in final state
IR safety important
T1/2,1
N partons in final state
18
Dijet Transverse Thrust Cross Section
Only statistical errors shown,sys is 10-15
O(as3) runsout here
Resummationmight helphere
19
Summary

• kT jet cross section suggests need to understand
  hadronization effects
• but will that be enough?
• Should we change the predictions or change the
  algo?
• Cone jet algorithm under scrutiny (CDF, DØ)
• Dijet Transverse Thrust analysis exhibits very
  clear need for better predictions (O(as4), Resum)