Progress in the cone analysis:Progress in the cone analysis:Optimization and systematic Optimization and systematic

checkschecks IntroductionIntroduction PerformancePerformance Systematic Systematic

checkschecks ConclusionsConclusions

IntroductionIntroduction PerformancePerformance Systematic Systematic

checkschecks ConclusionsConclusions

H. Ruiz (IFAE Barcelona), J. Nowell (Imperial College,London),

F. Teubert, A. Moutoussi (CERN)

ALEPH

IntroductionIntroduction FSI uncertainty largelarge and correlatedcorrelated between experiments. Effort has been put in trying to understand FSIunderstand FSI, with the aim of

reducing the systematic on mreducing the systematic on mWW.

In AprilApril, some promising results based on different cone algorithms were shown during a LEP-WW workshop.– A reduction of ~2 in CR error was achieved with a 20% loss in stats

Since then:– The algorithm has been optimized (moved to hybrid cones)

– Some new models have been tested.

– Systematic checks have been made.

Attacking FSIAttacking FSIIntroduction:Introduction:

Two different approaches:

1) Use of observables to put a limit in the effects:Use of observables to put a limit in the effects:• From Q analysis, BE effect is expected to be

drastically reduced (25MeV ~5MeV)

• For CR only extreme models discarded (sk1 high kI) second approach needed.

2) Redesign the analysis to make it less sensitive to FSIRedesign the analysis to make it less sensitive to FSI optimize jets algorithm.

• Durham chosen by optimizing just statistical error

Requirements on the new algorithm:– Efficient against the different CR models

– Minimum deviation from ‘standard’ analysis

The ‘new’ jet algorithm...The ‘new’ jet algorithm... Idea:Idea: FSI effects on mW come mainly

from inter-jet region, where– Ambiguity in clustering can occur.

– Momentum interchange is possible between particles from different Ws.

– Multiplicity variations are stronger (particle flow analysis, CR).

Proposed solution:– Apply a jet algorithm that excludes information from the

interjet region cone algorithm.

Price to pay: loss of statistical power.

Introduction:Introduction:

The The hybridhybrid cone algorithm cone algorithm The best performant cone algorithm tested is the hybrid:

– Take the particles of a given Durham jet.

– Find the cone of a given ‘radius’ that contains maximum momentum.

– Recompute direction of jet excluding particles outside the cone.

• Particles outside are used for energy computation

Only one parameter: opening angle R– R can be tuned to optimize stat and syst combination.

– In the limit of large R, ‘standard’ analysis is recovered.

Can be easily applied on Zs and semileptonic events for systematic checks.

R

Introduction:Introduction:

Performance against CR Performance against CR modelsmodels

standardreference: R=0.75

@ 189GeV

Statistical degradationStatistical degradationPerformance :Performance :

R=0.75

Performance with R=0.75Performance with R=0.75

mW shiftMODEL

Dur R=.75ratio

SK1 (100%) 276 168 1.64

SK1 (kI=0.65) 30 13 2.3

Hwg CR 33 18 1.8

Rathman CR 75 43 1.74

Stat error deterioration 1.13

Some details...Some details...

Energy fraction inside the cone:

Stat error vs energy fraction in cone:

R=0.75

Performance:Performance:

Higher energies...Higher energies...

For kI=0.65For kI=0.65

no clear trend shifts averaged

Statistical degradationStatistical degradation

No large variation

Performance:Performance:Impact in mImpact in mWW

Using data:

Durham Cone, R=0.75rad

Tot Stat Sys Tot Stat Sys

Sk1 66 51 41 64 56 31

Rath 95 51 79 75 56 51

Hwg 66 51 42 64 56 32

The systematic component of the error is strongly reduced with cones. In all scenarios the total error is smaller. The comparison gets better for the combination of the 4 experiments.

very very preliminary

MMWW vs E vs Ecm cm

100M

eV

200M

eVAverage:

very preliminary

Limits on kappa? Limits on kappa? (R=0.75)(R=0.75)

kIkI

Inte

gra

l

2

Results compatible to No CR within ~1sigma , preferance for some CRNo useful limit on kappa.

very preliminary

SystematicsSystematics

Cone algorithm may have different systematics than Durham.– Example: the angular distribution of EFs within a jet affects cone

and Durham in different ways.

Some checks done:

– Usual comparison Jetset-Herwig: mW~10MeV

(=standard).

– Apply cone analysis on semi-leptonic events.

– Systematic data-MC comparison of:

• A) EF distributions.

• B) Jet properties.

• C) Effect of cones.

Semileptonics for the Semileptonics for the proposed R=0.75proposed R=0.75

A) Angular distribution of EFsA) Angular distribution of EFsaround Durham jet axisaround Durham jet axis

Angle to jet axis (rad)

Energ

y

(arb

itr.

)

MC standardMC skiData

Another angle...Another angle... Not only angle to jet axis matters:

‘Azimutal symmetry’ of jet affected by: – jets nearby.

– CR?

closest jet other W

furthest jet other W

other jet same W

angle (rad)

……defining the angle...defining the angle...

angle to inter-jet plane

12

0 rad rad

Same W

1

2

closest jet other W

-angle for energy-angle for energy

Energ

y

(arb

itr.

)

all EFs

Adding momentum of all energy flows inside the cone (R=0.75)

MC standardData

angle (rad)

B) Jet propertiesB) Jet properties

Energy of the jets (GeV)

Durham

Cone

MC standardMC skiData

Azimuthal angleAzimuthal angle

Azimutal angle of the jets (rad)

Durham

Cone

MC standardMC skiData

Jets massJets mass

Jet mass (GeV)

Durham

Cone

MC standardMC skiData

C) Effect of cones: C) Effect of cones: momentum momentum kickkick

ppdur

pcon

e

p (GeV)

(rad)

MC standardMC skiData

… … zoom on zoom on kickskicks......

p (GeV)

(rad)

MC standardMC skiData

On ZsOn Zs

p (GeV)

(rad)

ppdur

pcon

e

MC standard

Data

Herwig

On ZsOn Zs

p (GeV)

(rad)

ppdur

pcon

e

MC standard

Data

Herwig

Angular bias Angular bias (on Z events)(on Z events)

Durham

R (rad)

R (rad)

Durham

Photons to charged:

All neutral to charged:

MC standard

Data

Angle between two components within a jet:

ConclusionsConclusions

The cone analysis:The cone analysis:– Reduces total mReduces total mWW error error

– Reduces strongly CR systematic yielding a more robust analysisReduces strongly CR systematic yielding a more robust analysis

No evidence for additional systematics from:No evidence for additional systematics from:– Jetset vs HerwigJetset vs Herwig

– Semi-leptonic testSemi-leptonic test

– Hadronic Ws and on ZsHadronic Ws and on Zs

Plans: fine tuning of R & more MC Models (Ariadne)Plans: fine tuning of R & more MC Models (Ariadne)