Jorgen D’HondtJorgen D’HondtUniversity of BrusselsUniversity of Brussels

Short reminder about the analysis Short reminder about the analysis ‘ ‘Final’ results of full LEP2 data setFinal’ results of full LEP2 data set Systematic uncertainties : Colour Systematic uncertainties : Colour ReconnectionReconnection LEP WW Workshop (February 2003)LEP WW Workshop (February 2003)

DELPHI General Meeting, CERN, March 2003

…or some publicity for my thesis

W

W

lepton

neutrino

quark

anti-quark

Possible WW final state

• qq’QQ’ : fully hadronic (BR~45%)

• qqll : semi leptonic (BR~44%)

• lll’l’ : fully leptonic (BR~11%)

BR = 1/3

BR = 2/3

• CC03 W-pair production• Born or tree level

• Separation of W-vertices ~0.1 fm (due to small range of weak force)• Independent W decay ??

Cross talk systematics

Up to 40000 WW events selected at LEP2Up to 40000 WW events selected at LEP2

MW reconstructed from decay products of W boson MW resolution : ~10 GeV~10 GeV measured and ~3 GeV~3 GeV after kin.fit

qqQQqqQQ

Kinematic fit :Kinematic fit : Conservation of (p,E) : Conservation of (p,E) : 4C4C Improves E resolutionImproves E resolution

Analysis methods :Analysis methods : Reweighting Reweighting A,L,OA,L,O Convolution Convolution DD

Challenge :Challenge : Jet pairingJet pairing Hard gluon radiationHard gluon radiation

detaildetail

before 4C fit

after 4C fit

Full data except TPC-S6

TPC-S6

Very nice plots for the publicationI consider them as final !!

QFTQFT

e-

e+

object 1

object 2

initial stateinitial state final statefinal state

jetpairing ambiguityclustering ambiguityISR ambiguity

4C : conservation of (E, p)[Ee-+ Ee+ , pe-+ pe+ = 0]initial [Eobj.1+ Eobj.2 , pobj.1+ pobj.2 ]final

E and p resolution parametrized from jet-angles (DELPHI)

2i (m1,m2)

All jet pairings weight(r) (cos W- , Q(W-,W+) , 1/kt gluon jet)

3 clustering algorithms weight(s) = (45% , 36% , 19%)

ISR hypothesis lost in beam pipe 4C 3C kin.fit

if pz/pz 1.5 sigma

MC parameterized weight wISR pz/pz

wISR.(3C) + (1-wISR).(4C)

MCMC

mW+

mW-

+

r {1,…3,…,10}

s {1,2,3}

t {1,2}

pi(mW+,mW-) = wr,s,t . pir,s,t(mW+,mW-)

r,s,t

if 5 jet eventpi(mW+,mW-) exp[-0.5 i

2(mW+,mW-) ]

reflects the reconstructed mass information from the kinematics of the event

Construction of 2D experimental ideogram Construction of 2D experimental ideogram ppii(m(mW+W+,m,mW-W-))“ likelihood ratio functionlikelihood ratio function “

Th.pdf(mW+,mW-,MW) = Peff . Signal (mW+,mW-,MW) + (1-Peff) . Background

PS

MC event purity from jet topology (jj,Ej)WW and ZZ Breit-Wigner terms and Phase-Space correction

Li(MW) = L(MW) maximum likelihood MMWWi{events} ( calibration needed )

Construction of event likelihood Construction of event likelihood LLii(M(MWW))

“convolution of theoretical predicted function and ideogram convolution of theoretical predicted function and ideogram ”

• TPC-S6TPC-S6 period not treated different for mass but for the width the shift was applied.• The 2/ndf of the fits are acceptable and are propagated into the systematic uncertainties

• Slope was linear and compatible with unity for mass estimator.• Reweighting method as been successfully checked !! (W width)

2.6 test2.6 test2.092.09

Generated W widthGenerated W width

BiasBias

• Complete data set except TPC-S6TPC-S6 period• Good agreement between expectation and data values...

W mass W width

symmetric uncertainties asymmetric uncertainties

• ~4 bug for width was mine… is fixed now !!!• Good agreement between expectation and data values…

QFTQFT

ee--

ee++WW++

WW-- PSPS

PSPS

LEP beam energy

fragmentation modeling *

cross-talk *between W’s

ISR

illustration of main systematic uncertaintiesillustration of main systematic uncertainties

* processes not known from first principles

Method has been verified to high precision (~15%)

Jet reconstruction systematics !!

Hadronic WW channel within DELPHI

ref

The total shift can be downscaled according to the LEP combined measurements. For the moment ~25% fraction of the LUBOEI BE32 model.

4040 1010 MeV/c2 (W mass)

largest systematic uncertaintylargest systematic uncertainty

……but !!but !!

W decay vert. ~ 0.10.1 fm distance hadronisation scale ~ 11 fm

large space-time overlap

suppressed in parton shower :

due to group structure of QCD at least 2 gluons must be emitted

W width restrict the energy range of primary gluons from q1Q4 and q2Q3

~ (s2/MW) W/Nc

2

CR effect could occur in the confined region ???

s<<1 s1

Models to emulate the CR effect

T.Sjöstrand and V.A.Khoze, Z.Phys. C62 (1994)

PYTHIA : string reconfiguration if they overlap or cross in space-time

– SK1SK1 (lateral flux tube) : via event string overlap O : ppCRCR = 1-exp(- = 1-exp(-·O)·O)

ARIADNE : rearrangement of colour dipoles to reduce the string length (mass)

– AR2 : only after soft gluon radiation (Eg < W) HERWIG : rearrangement of colour dipoles changing the size of the clusters

CR implemented in existing fragmentation models :

SK1 mSK1 mW W shiftshift

PPCRCR

KK ~ 0.66 ~ 0.66

Latest preliminary predictions for the W-mass• PYTHIA (SK1) ~ 5050 MeV/c2

• ARIADNE (AR2) ~ 7070 MeV /c2

• HERWIG ~ 4040 MeV/c2

Statistical uncertainty (LEP2) ~ 3030 MeV /c2

CR

Design observable sensitive to the CR effect :

particle flow (counting particles RR)

RR =

2.6 2.1 49 %

Not sensitive to our observable ????A + B

AB

C

D

C + D

mW is assumed to be unknown but it must be invariant for different

estimators

mmWW(std)-m(std)-mWW(cone)(cone)

from MC as function of

Correlation between W mass estimators ~ 83 %

uncertainty on the difference is small

Influence on W mass estimator (MeV/c2) for different values of (SK1SK1)

MC predictionMC prediction

DELPHI preliminary

no correlation assumedno correlation assumedbetween the twobetween the two

measurement methodsmeasurement methods(good approximation)(good approximation)

combination particle flowparticle flow (LEP) and mmWW (DELPHI)

Updated DELPHI measurementof mmWW for Moriond Conference

~100 MeV/c2

no correlation assumedno correlation assumedbetween the twobetween the two

measurement methodsmeasurement methods(good approximation)(good approximation)

68% CL for [[0.78 , 2.180.78 , 2.18]]

central value1.371.37

LEP expected :

70 MeV/c2 106 MeV/c2 144 MeV/c2

General case, no calibration, DELPHI only

DELPHI all channels (preliminary)

Most important point was the ‘bug’ in Most important point was the ‘bug’ in the ALEPH Monte Carlo simulation of the the ALEPH Monte Carlo simulation of the energy clusters in their calorimetersenergy clusters in their calorimeters Effect on LEP W mass (not official and Effect on LEP W mass (not official and not to be made public !!!!!)not to be made public !!!!!)

DELPHI General Meeting, CERN, March 2003

February 2003

Apply momentum cuton particles entering theW mass analysis

evqq vqq vqq

qqQQ lvqq

Results relative to standardmomentum cut in theparticle selection

In qqQQ channel sensitiveto Colour Reconnectionbut this cannot be in thelvqq channel

Shifts in GeV

July 2002

Summary of their bugs :• forgot to include the magnetic field in their simulation of the calorimetersforgot to include the magnetic field in their simulation of the calorimeters include it !!• remnants of EM showers or satellites of electron not perfect similated (type 6)remnants of EM showers or satellites of electron not perfect similated (type 6) ECAL cleaning, remove the remnants in data and MC !!

… … all similar to DELPHI, but ...all similar to DELPHI, but ...

Bhabha events at Z peak Bhabha events at Z peak

ECAL HCAL

Correct B field

no B field

Summary of their bugs :• forgot to include the magnetic field in their simulation of the calorimeters include it !!• remnants of EM showers or satellites of electron not perfect similated (type 6)remnants of EM showers or satellites of electron not perfect similated (type 6) ECAL cleaning, remove the remnants in data and MC !!

… … all similar to DELPHI, but ...all similar to DELPHI, but ...

Used to cut away anything within 2.5 degrees from track

Now increased to 8 degree

ECAL barrel region

Correct B field

no B field

cut

moved

evqq vqqvqq

qqQQ lvqq

Shift between newand old mW value

• Including B field• Including ECAL cleaning

evqq vqqvqq

In the LEP combined W mass ALEPH has a very large width

~46%(due to small systematic errors)

Not to be made public outside of this room !!!Not to be made public outside of this room !!!

LEP combined / all channels

mW = 80.446 ± 0.042 80.411 ± 0.043

direction of shift

~10 GeV

Mixed Lorentz Boosted Z0’s !!!!Other experiments didn’t investigated this technique.

Only qqQQ channel, to be checked in lvqq channel.

Hadronic W mass/width analysis readyHadronic W mass/width analysis ready All systematic uncertainties estimatesAll systematic uncertainties estimates Some of them are surprisingly large Some of them are surprisingly large ((electroweak radiative correctionselectroweak radiative corrections) and ) and not not completely understoodcompletely understood Some can have a more optimal treatment Some can have a more optimal treatment ((Colour ReconnectionColour Reconnection) but should be ) but should be desided desided together with other LEP experimentstogether with other LEP experiments Measurement of Colour Reconnection is Measurement of Colour Reconnection is very very importantimportant

… … ALEPH will have new updated resultsALEPH will have new updated results … … indirect Higgs mass will shift by +10 indirect Higgs mass will shift by +10 GeV/cGeV/c22