Experience with muons in analysis : W’ search

Experience with muons in analysis:Experience with muons in analysis:W’ searchW’ search

Carmen Diez Pardos

CIEMAT Madrid

Muon Barrel Workshop

Physics sesion

25/02/2011

1

OutlineOutline

Introduction Muons in W´ searches

Analysis concept Efficiency determination in W´ (and W

xsec measurement) Background determination Systematics Results

Outlook

24/02/2011 MB Workshop C. Diez Pardos (CIEMAT) 2


1. Introduction: about this talk1. Introduction: about this talk

The Wʹ→μν Search at the LHC• Jump in collision energy (from TeVatron to LHC) single most important factor in search for new heavy particles

• (Relatively) easy analysis• High-pT (unprescaled) single muon trigger • Reconstruction of MT by combing high-pT muon, MET • Practically background-free channel

Study muon efficiencies, effect of muon momentum resolution and scale, at the moment without high pT muons from collisions

An analysis that can lead to one of the first LHC discoveries!

Present an example of a physics channel focusing on the aspects related to muons, as efficiency measurements, reconstruction and systematics

24/02/2011 MB Workshop 4

Spectacular signature• One high-pT muon in event (“straight track”)• Plus, “nothing else”

Experimental limits from Tevatron in the channel W‘ → ev: m(W‘)>1.1 TeV [CDF, published]

CMS limit from electron channel m(W‘)>1.36 TeV

1 Muon with ~550 GeV

Wʹ→Wʹ→μνμν Topology and Limits Topology and Limits

C. Diez Pardos (CIEMAT)

Event with Max MT


Analysis strategy Analysis strategy • Identify high-pT (unprescaled) single muon trigger

•Follow Exotica Muon group recommendation for muon selection (Backup slides)

• Reconstruction of MT by combing high-pT muon, MET

• Determine efficiencies: Efficiencies used for our signal MC (the BG is obtained from data) and to crosscheck if we understand our BG from data/MC comparisons

• Determine BG from pp collisions (fit in MT sideband extrapolating to signal region) and cosmics

• Comparison between data and MC MT distributions: No excess seen in data Set a limit

Samples: DATA: 36/pb Nov4ReReco, MC BG: Fall10 38X samples


1. Object reconstruction:1. Object reconstruction:MUONS: Extensive studies of high-pT muon reconstructor

• Global, tracker-only, TPFMS, cocktail, picky, TMR, DYT: Using „cocktail“ as recommended by Exotica Muon group

• For this analysis tails of the pT distribution matter

• Using both MC and cosmics/CRAFT (see also Z`→ note)

MET reconstruction: For the dominant background MET is driven by muon-pT measurement: hadronic component less relevant for tail of MET (MT) distribution


Why are they important? For BG – comparison data/MC to extract systematics, differences could indicate the source of disagreement For the signal: used to extract the limit (or the xsec!)

There are no high-pt muons in collision data, in order not to rely just on MC we use

A data driven method: Tag and Probe using Zmumu events (AN-2010/233): not the official TP package, different background subtraction, no analytical description of the Z->mumu shape For trigger efficiencies, also JetMet triggered data (get to higher

values of pT)

The agreement at medium pT between data and MC is also seen at high pT with cosmic muon data.

2. Efficiency2. Efficiency

measurementmeasurement


Single muon id. efficiency: T&P methodSingle muon id. efficiency: T&P methodSelection of a Z->mumu sampleSelection of a Z->mumu sample

Preselection of tracks with kinematics cut pt >25. GeV and |eta| < 2.1which combined with an opposite charged track give an invariant mass inthe range 60. < Mμμ < 120. GeV.Match track-muon is done with the track reference in the muon object.Tag muons must fullfill the following selection: (Exotica, VBTF recommendation)

Must be Global and Tracker MuonsCombined isolation < 0.15 in a cone R < 0.3Quality cuts related to the track: χ2 < 10/ndof, d0<0.02 cm, number of valid muon and pixel hits >0, number of valid tracker hits >10, number of matching segments >1, valid muon hits >0Muon matched to a L3 muon:

hltSingleMu9L3Filtered9 (RunA + Run B until 147196) hltSingleMu15L3Filtered15 (till run 149442)

The other track is considered as a probe to study the efficiency.


2.1 Trigger Efficiencies2.1 Trigger Efficiencies• Mixture of triggers in the data, with raising pT threshold • Always use lowest-threshold un-prescaled single-muon trigger: HLT_Mu9, HLT_Mu11, HLT_Mu15

All trigger threshold lower than the pT>25 GeV requirement

•Trigger efficiencies determined with tag-and-probe

Flat for pT>25 GeV

flat trigger efficiency is used throughout the analysis

Run 2010B


2.1 Trigger Efficiencies2.1 Trigger Efficiencies

Run2010A0.854+-0.010

Run2010B0.914+-0.010

Run2010B0.913+-0.010

Trigger efficiencies in data are determined with a complementary method (1.5% difference): using jet triggered samples with offline muons, can reach higher values of pT

In data no muons with pT>240 GeV while muons from potential W‘ signal are O(>=500 GeV)

Extrapolation for high pT needed, done with MC

→ reasonable agreement with the data driven studies for low pT

Efficiency from

Data ~

92%

11

2.2 Data: T&P Determined 2.2 Data: T&P Determined EfficienciesEfficiencies

Excellent agreement between MC-predicted and data-derived (T&P) efficiencies (largest discrepancy coming from trigger: 4% assigned as systematic uncertainty)

εMu

Muon reconstruction

εSEL Selection

εISO Isolation

εTRK

Reconstruction inner track

Efficiencies are only used for our signal MC, the background is obtained from data

24/02/2011 MB Workshop C. Diez Pardos (CIEMAT)


2.2 Data: T&P Determined 2.2 Data: T&P Determined EfficienciesEfficiencies

Tracking Isolation

Selection Reconstruction

Data

MC

Other Efficiencies to High Other Efficiencies to High pTpT



(An analysis with signal…)(An analysis with signal…)Efficiencies for W cross-section measurementEfficiencies for W cross-section measurement

Same method used for the W->munu cross-section measurement with all 2010 data (already pre-approved) Advantage of extracting the efficiencies on a sample of muons kinematically very similar

Data

MC

The selection is slightly different (d0< 0.2 cm iso<0.1)


W cross-section measurementW cross-section measurement

The ratio is roughly flat in pT, the correction is applied in eta bins Good agreement between data and MC

MCdata

LdtA

NNWBr

W

bckgobs

W

)(

The correction factor is incorporated to the cross-section measurement, as a correction to Aw*eff factors


Uncertainty on the number of background eventsdue to statistical uncertainty of the sideband fit itself

Signal

Background

Source of systematic uncertainty Value

Luminosity 11 %

Muon pT resolution 0.14 TeV-1

Muon momentum scale 0.4 %

MET resolution 10 %

Trigger and other efficiency 4 %

(Back to W’)(Back to W’)4. Systematic uncertainties4. Systematic uncertainties

Focus on systematics related to muon

Systematic in MT due to momentum Systematic in MT due to momentum scale and resolutionscale and resolution

Method: 'Distort' the muon momentum by certain amount, recreate MET with the new momentum and recalculate MT.

Assign the relative difference between # of events with 'standard' MT and 'distorted MT”> 600 GeV as a systematic

Muon distorted like: Momentum scale pt'=pt+a0*pt^2

Momentum resolution 1/pt' = 1/pt + C*Gaussian(0,1)

24/02/2011 MB Workshop 17C. Diez Pardos (CIEMAT)

Updated studies with all data, effects also depend on eta, phi

Systematic in MSystematic in MTT due to momentum due to momentum

scale and resolutionscale and resolution Realistic values for the resolution and momentum scale from data (results available at the time of doing the

analysis, most conservative values taken, it doesn´t affect significatively the analysis) MuscleFit (CMS AN-2010/059), SIDRA (CMS AN-2010/059) studies for Z->mumu – intermediate pT (a0 =

0.0039/40GeV, C = 0.14/TeV)

Cosmic studies: alignment, end point analysis (pt' = pt + ( 0.0031 +/-0.0005)*pt, C = -0.047/TeV) (CMS PAS TRK-10-004, http://indico.cern.ch/conferenceDisplay.py?confId=101503)

For W cross-section this systematics (with the latest results from muon momentum resolution and scale) is estimated to be <0.1%

Results

Cosmics are the only way to have information about high-pT (TeV) muons

24/02/2011 MB Workshop 18C. Diez Pardos (CIEMAT)

Updated studies with all data: http://indico.cern.ch/getFile.py/access?contribId=5&resId=0&materialId=slides&confId=127861


Search strategy & limit setting: good agreement data-MC, no excess seen → set an exclusion limit

Apply search window for high MT region and compare #expected - #observed events

Use a sliding search window to optimize the limit as done for electron channel

Search window

Combined exclusion m(W‘)<1.58 TeV

Present Tevatron limits m(W‘)<1TeV [D0 publ.], 1.1TeV [CDF, to be pulished]

Final results: Final results: MT Distribution & Limit setting

OutlookOutlookSearch for W´ performed with 36/pb

No excess seen in data set exclusion on W´-mass with SM-like couplings of m(W‘)<1.39 TeV, combined with electron exclusion: m(W‘)<1.58 TeV

Trigger, reconstruction and selection efficiencies studied in data (up to pT~200 GeV) with tag-and-probe and extrapolated to higher pT

Cosmics are still our only high pt muons...

Used cosmic data to cross-check the muon momentum resolution, scale from Z->mumu studies

Next year challenge: start dealing with real high pT muons from collisions


Back up Back up



Samples: Data and MCSamples: Data and MCIm thinking about removing the tables or at least some columns,and summarise it in a couple of lines

DATA: 36/pb Nov4ReRecoMC BG: Fall10 38X samples

MC Signal


Sideband

Get total pp background from MT data spectrum

Choose a region with low signal contamination (~1%)

Fit sideband, extrapolate to high MT tail to predict # of (pp) background events in signal region

Method proven to work in MC

• Background in MC is sum of all SM contributions (but mainly: W→μν)

Cosmics not described by this method

Extrapolation region

Method

(Back to W’)(Back to W’)3. Background determination3. Background determination


•Goal: determine potential cosmics contamination, not described by sideband fit

•Cosmic contamination more relevant for tail of MT distribution

Cosmic contamination for d0 < 0.02 cm

pp

cosmics

3.1 Cosmic Background3.1 Cosmic Background

Table for d0 OptimizationTable for d0 Optimization


Backup: Search Window for Backup: Search Window for C&CC&C


Cut-and-count method


1. Select a muon: Trigger selection: HLT_Mu9, HLT_Mu11, HLT_Mu15

2. Muon identification and quality cuts: Follow Exotica Muon group recommondation

◦ Tracker & global muon

◦ ≥ 11 silicon tracker hits

◦ ≥ 1 pixel hit

◦ χ² /Ndof < 10

◦ Transverse impact parameter d0 < 0.02 cm(order of magnitude tighter than VBTF recommondation)

◦ At least 2 matched muon segments3. Acceptance η < 2.1

4. Exactly 1 global muon with pT > 25 GeV

5. Relative combined isolation in a cone (ΔR < 0.3) < 0.156. Dedicated kinematic selection

◦ 0.4 < pT / MET < 1.5

◦ Δ(φμ,υ) > 2.5

SelectionSelection

More on T&P methodMore on T&P method






OutlookOutlook Search for W´ performed with 36/pb

No excess seen in data set exclusion on W´-mass with SM-like couplings of m(W‘)<1.39 TeV, combined with electron channel combined limit m(W‘)>1.58 TeV

Trigger, reconstruction and selection efficiencies studied in data (up to pT~200 GeV) with tag-and-probe and extrapolated to higher pT

Cosmics are still our only high pt muons...

Main background Standard Model W, others contribute <10% in region high MT. Data-driven method for background: using sideband-fit to determine total background in signal region Check the effect of muon reco, etc

Next year: start dealing with real high pT muons CHALLENGE, NEED FOR COSMICS???


Documents

Experience with muons in analysis : W’ search