Summary of Higgs session: Experimental part Ia Iashvili SUNY at Buffalo

Summary of Higgs session:Summary of Higgs session:

Experimental partExperimental part

Ia IashviliIa Iashvili

SUNY at BuffaloSUNY at Buffalo

For the Higgs Working groupFor the Higgs Working group

TeV4LHC Workshop Fermilab, 22 October 2005

OutlineOutlineTevatron Higgs searchesTevatron Higgs searches Experience gainedExperience gained What can Tevatron achieve before LHCWhat can Tevatron achieve before LHC

Example of Tevatron data being used for testing LHC predictionsExample of Tevatron data being used for testing LHC predictions

Some LHC Higgs studiesSome LHC Higgs studies

Summary and outlookSummary and outlook

Only selected topics. Contributions at earlier TeV4LHC meetings won’t be Only selected topics. Contributions at earlier TeV4LHC meetings won’t be covered. These will be documented in the workshop write-up.covered. These will be documented in the workshop write-up.

2

Introduction

ggH

HW

HZHqq

HttHbb

(pp H + X) [pb]√s = 2 TeV

MH [GeV]H bbH WW(*)

Dominant decay modes

Goal:Goal:

Achieve aboveAchieve above

Help to achieve this Help to achieve this (similar (similar sensitivity in Atlas)sensitivity in Atlas)

3

WHlbb searches at Tevatron

WH/ZH, Hbb is best for light Higgs searchAt MH=115GeV

ZHnnbb … 15 events/fb-1 ZHllbb … 2 events/fb-1 WHlnbb … 14 events/fb–1

WHlbb: high pT isolated lepton, missing ET and two b-jets

non-W QCDfalse isolated leptons or

false missing energy

top quark production

physics background

mistags in W events

false b-tags with

lepton + MET

W + heavy flavor

physics background

Each background estimate is a miniature analysis unto itself.Techniques can be spun off to measure other physics processes.

BackgroundBackground

Should be estimated entirely from dataShould be estimated entirely from data

Strong motivation for measuring top production cross section precisely!

Mistagging rate estimated in dataMistagging rate estimated in data

Flavor composition from MC (Alpgen) Flavor composition from MC (Alpgen) and cross checked in dataand cross checked in data

4

WHlbb searches at Tevatron

Check efficiency in dataevents vs. efficiency in simulation:need scale factor of 0.91±0.06

B-tagging performanceB-tagging performance

Checking event counts Checking event kinematics

5

WHlbb searches at Tevatron

6

ZHbb searches at Tevatron

110o

Missing ETb-jet

b-jet

120o

y

x Signal has a distinctive topology Large missing transverse energy two high pT b-jets No isolated leptonsJets are acoplanar

But diffical to trigger – no high-pT lepton in the event trigger on missing ET and jets

Suffers from large background“physics” backgrounds W+jets, Z+jets, top, ZZ, and WZ Can be estimated from MC“instrumental” backgrounds QCD multijet events with mismeasurement

of jets Estimated from data

Need to find smart variable to separate signal from background

7

ZHbb searches at Tevatron

For the 120 GeV Higgs mass in ±20 GeV mass window around the expected reconstructed peak value (Dijet mass resolution is ~17 %):

SM background prediction: 4.36 1.02 events QCD (11.4%), Top (20.5%), EWK (18.2%), Light flavor mistag(50%)Observed: 6 events.

Control Region 2 – EWK Require min. 1 lepton Missing ET and 2nd leading jets are not parallel Optimized cuts are tested in this region before looking at the real data in the Signal Region

Control Region 1 – QCD h.f. Veto events with identified leptons Missing ET and 2nd leading jet are parallel

Extended Signal Region (no optimization) Veto events with leptons Missing ET and 2nd leading jet are not parallel Cut optimization is performed in this region based on MC simulation before looking at the data

8

ZHbb searches at Tevatron

Mass

Window

105GeV

[70,120]

115GeV

[80,130]

125GeV

[90,140]

135GeV

[100,150]

Data 4 3 2 2

Acc (%) 0.29 0.07

0.33 0.08

0.35 0.09

0.34 0.09

Total BKG 2.75 0.88

2.19 0.72

1.93 0.66

1.71 0.57

sideband sidebandsignalPhysics bkgd from MC

Instrumental bkgd from sidebandsAsym = (ET-HT)/(ET+HT)

Rtrk = |PTtrk-PT,2trk|/PTtrk

Wjj/Wbb 32%, Zjj/Zbb 31%, Instr. 16%,Top 15%, WZ/ZZ 6%

9

HW+W-ll searches at Tevatron

Two high pT isolated lepton and large missing ET and no hard jets Clean and easily triggerable signatureClean and easily triggerable signature Sensitive to tau channels with leptonic decaySensitive to tau channels with leptonic decay

One of the most promissing channels at LHC as well

Backgrounds:WW, WZ, ZZ, DY, ttbar estimated from PYTHIA and normalized NLO cross section calculations.W+jets(jete/m) estimated (at least partially) from datamultijets events estimated from data

Efficiencies:Efficiencies:Lepton triggering, reconstruction and Lepton triggering, reconstruction and identification efficiencies all have to be identification efficiencies all have to be determined in data and factorized in MC for determined in data and factorized in MC for signal rate estimation signal rate estimation Precise estimation is importrant since no Precise estimation is importrant since no “bump” can be observed from Higgs“bump” can be observed from Higgs

WW production is the dominant background compulsory to first measure the WW production cross-section

[PRL 94, 211801 (2005)] (CDF)

(W(W++WW--)=13.8 )=13.8 +4.3+4.3(stat)(stat)+1.2+1.2 (sys)±0.9(lumi) pb (sys)±0.9(lumi) pb-3.8 -0.9-3.8 -0.9

(W(W++WW--)=14.6 )=14.6 +5.8+5.8(stat)(stat)+1.8+1.8 (sys)±0.9(lumi) pb (sys)±0.9(lumi) pb-5.1 -3.0-5.1 -3.0

[PRL 94, 151801 (2005)] (DØ)

Consistent with thepry prediction of12-13.5 Consistent with thepry prediction of12-13.5 pb (J.Ohnemus, J.M. Campbell, R.K.Ellis)pb (J.Ohnemus, J.M. Campbell, R.K.Ellis)

10

HW+W-ll searches at Tevatron

W-

W+e+

e-

Exploit spin correlations

Leptons tend to be parallel -- small (ℓ, ℓ) Neutrinos go parallel -- typically larger missing energy than WWSmall di-lepton invariant mass

11

Tevatron SM Higgs Sensitivity: expectations two years ago

Prospects updated in 2003 in the low Higgs mass region W(Z) H l(,ll) bb

better detector understanding optimization of analysis

Sensitivity in the mass region above LEP limit (114.4 GeV ) starts at ~2 fb-1

With 8 fb-1: exclusion 115-135 GeV & 145-180 GeV, 5 - 3 sigma discovery/evidence @ 115 – 130 GeV Meanwhile understanding detectors better, optimizing analysis techniques measuring SM backgrounds (Zb, WW, Wbb) Placing first Higgs limits which can be compared to the prospects

12

Sensitivity with existing Tevatron analyses

Cross-Section times branching fraction limitas a multiple of the SM rate

We should be around 6 at low masses, not around 12-20 with the current lumi (0.3 fb-1).

Where can we gain ?

CDFCDF DØDØ

the “kink” at around 140 GeV goes away

Work in progressWork in progress Work in progressWork in progress

13

So how do we get there?

Step 1 (for early 2006) WH/ZH:

Optimize b-tagging (Looser) Combine single and double tag S/sqrt(B) is 40-50% in single tag compared to double tag. Equivalent to 20% more lumi than double tag alone.

WH(e): include Phi-cracks WH(): combine single- and +jets trigger ZH : optimize Selection

DØDØ

Step 2 WH/ZH:

include WHWWW and Z l+l- channel ! (*1.3)use Neural Net Tagger (*1.34*1.34)use Neural Net Selection (*1.8) use TrackCalJets mass resolution (*1.3)

WH(e): include End-Cap calorimeter WH(): improve QCD rejection loosen b-tagWH : include W (*1.4)

All these improvements will All these improvements will bring us to the expected level bring us to the expected level of sensitivityof sensitivity

14

So how do we get there?

Improvement WHlbb ZHbb ZHllbb

Mass resolution 1.7 1.7 1.7

Continuous b-tag (NN) 1.5 1.5 1.5

Forward b-tag 1.1 1.1 1.1

Forward leptons 1.3 1.0 1.6

Track-only leptons 1.4 1.0 1.6

NN Selection 1.75 1.75 1.0

WH signal in ZH 1.0 2.7 1.0

Product of above 8.9 13.3 7.2

CDF+DØ combination 2.0 2.0 2.0

All combined 17.8 26.6 14.4

Start with existing channels, add in ideaswith latest knowledgeof how well they work.

Expect a factor of ~10 luminosity improvement perchannel, and a factor of 2 from CDF+DØ Combination

CDFCDF

15

Expected Signal Significance CDF+DØ vs Luminosity

per experiment per experiment

mH=115 GeV assumed

Work in progressWork in progress

Work in progressWork in progress

16

Improving Jet Energy Resolution Using Tracks

Idea Reconstruct calorimeter-based jets (0.5 cone) Use track momentum measurements to set an accurate scale for hadron response for each hadron in the jet.

Proposed in CMSCMS Note 2004/015, O.Kodolova et al. “Jet energy correction with charged particle tracks in CMS”

Tevatron provides an excellent opportunity to test and optimize this technique on real collider data.

Out-Jet

In-Jet

Propagate tracks to the calorimeter surface. dca(xy) < 0.5cm, dca(z) < 1.0 cm.

Classify tracks: R(vtx)<0.5, R(cal)<0.5 : IN jet R(vtx)<0.5, R(cal)>0.5 : Out-of-cone

For each IN-jet track: Etrkjet=Ecaljet +(1-F)Etrk

For each Out-of-cone track: Etrkjet=Ecaljet +Etrk

where F is a single pion calorimeter response

17

Improving Jet Energy Resolution Using Tracks

Performance in Performance in ZZbbbar MC bbbar MC

eventsevents

Performance in Performance in +jet data events+jet data events

12% improvement in mass resolution.12% 20% by optimization of the TrackCalJet algorithm

10-20% jet resolution improvementin data at 40 GeV.Higher improvement at lower pT.

In CMS MC studies: ~40% improvementat the same energies

18

QCD Higher Order Corrections in H + 1jet at the LHCLow mass Higgs searches with H in association with high PT jets are crucial at the LHC

NLO QCD corrections for VBF signal and Z+jets in H+2jet analysis have been considered in the past

QCD Higher order corrections have not been

evaluated within the H+1jet analysis neither for signal nor for the Z+jets background

NLO corrections are evaluated here with MCFMAlso address the impact of Z+2-3jet tree level ME on Z+jets using ALPGEN/SHERPA

QCD HO corrections are large in the

region of the phase space where the signal-to-background is optimal for searchesQCD Z+1j is enhanced by a factor of 2 Signal, H+1j is enhanced by a factor 1.75Need to re-optimize the analysisSignal significance does not decrease

Tag jet

Not Tagged Large PTH & MHJTag jet

19

Summary and outlook

Much has been learned from Tevatron Higgs searches on various challenges and issues faced at hadron collider environment

We have also learned how important current experience is to actually achieve expected performance

Tevatron will provide important information on Higgs sector before LHC. For low mass Higgs searched Tevatron is complementary to LHC.

We are able to test LHC predictions using Tevatron data and provide important feadback

The plan is to document our experience and findings in the Workshop proceeding (beginning of next year)