The CMS detector Two-particle correlation functions Results in m inimum bias collisions

Zimányi 2010 Winter School on Heavy Ion Physics, 29 Nov 2010, RMKI/ELTE Budapest 1Gábor Veres

• The CMS detector

• Two-particle correlation functions

• Results in minimum bias collisions

• Results in high multiplicity collisions

• Cross-checks

Gábor I. VeresCERN Geneva and ELTE Budapest

CMS Collaboration

Based on:- JHEP 1009:091,2010- My talk at ECT* Trento: QCD at the LHC, September 28, 2010- CERN seminar talks by Gunther Roland and Guido Tonelli, Sept. 21, 2010

Zimányi 2010 Winter School on Heavy Ion Physics, 29 Nov 2010, RMKI/ELTE Budapest



The CMS Silicon Tracker

• Coverage up to ||<2.5; extremely high granularity, to keep low occupancy (~ a few%) also at LHC nominal luminosity.

• Largest Silicon Tracker ever built: Strips: 9.3M channels; Pixels: 66M channels. Operational fractions: strips 98.1%; pixel 98.3%

TOBTOB

TIDTIDTIBTIB

TECTEC

PDPD

PDPD

TIBTIB

TOBTOB


Angular Correlation Functions

I. DefinitionCorrelation Functions:

II. Anatomy


Correlation Function Definition

Background distribution:

BN (,) 1N 2

d 2N bkg

dd

R(,) (N 1)SN (,)BN (,)

1

N

pT-inclusive two-particle angular correlations in minimum bias collisions

1 2

1 2

SN (,) 1N (N 1)

d 2N signal

dd

Signal distribution:

Same event pairs Mixed event pairs

CMS pp 7TeV

Ratio Signal/Background


Angular Correlation Functions

Short-range correlations ( < 2):Resonances, string fragmentation,“clusters”

Bose-Einstein correlations: ~ (0,0)

CMS 7TeV pp min bias

“Near-side” ~ 0) jet peak:Correlation of particles

within a single jet

“Away-side” ~ ) jet correlations:

Correlation of particles between back-to-back jets

Momentum conservation:~ -cos()


Correlations in Min Bias pp

CMS pp Data

Pythia D6T


Short-Range Correlations vs. s

1D “Projection” to axis

PYTHIA describes the energy dependence,

matches cluster width in data,but underestimates the cluster multiplicity Keff

Keff: Number of correlated particles: correlation width in

CMS

CMS


High Multiplicity Events

268 reconstructed particles in the tracker in a single pp collision:the highest multiplicity event in ~70 billion inelastic events sampled (1/pb)


Why study extreme multiplicities?

Our most recent correlation studies focus on the tail of the distribution, where several MC generators severely under-estimate the data(an exception: PYTHIA8).

Motivations: Trying to find (more) unexpected effects in this regime Learn more about (soft) QCD and particle production mechanisms with more differential measurements Highest multiplicities in pp begin to approach those in ion collisions; can we learn something about similarities or differences?


High Multiplicity Trigger

Level-1 (hardware):Requires ET> 60 GeV in calorimeters

High-Level trigger (software):More than 70 (85) tracks with pT > 0.4 GeV/c, || < 2, within dz < 0.12 cm of a single vertex with z < 10 cm.~50% CPU usage of the HLT

Dedicated trigger was needed to record highest multiplicities


High Multiplicity Event Statistics

Two different HLT thresholds: Nonline > 70 and Nonline > 85

HLT85 trigger range un-prescaledfor full 980nb-1

out of 5x1010 collisions

Multiplicity binning usespT > 0.4 GeV/c

|| < 2.4

1000 times morehigh multiplicity eventsrecorded with the triggercompared to Min. Bias


Event and Track Selection

Event-selection and analysis done with tracks pointing to primary vertex with O(100m) resolution


Results : data, inclusive pT

MinBias high multiplicity (N>110)

Jet peak/away-side correlations enhanced in high multiplicity eventsAbundant jet production in high multiplicity sample



Results: data, inclusive pT

After cuting off the jet peak at (0,0) we can observe:Structure of away-side ridge (back-to-back jets)

Small change for large around ~ 0 ?


Results: data, pT: 1-3 GeV/c

Pronounced new structure at large around ~ 0 !


CMS Collab., JHEP 1009:091,2010, arXiv:1009:4122


Illustration of the effect

Particles surfacing in the same time zone, but far away in latitude, talk to each other…

…What mechanism is

the “telephone line”??!

p

p


Correlations in PYTHIA8

No structure at large Same for Herwig++, madgraph, PYTHIA6


Multiplicity- and pT -Dependence

“Ridge” maximal for highest multiplicity and 1 < pT < 3 GeV/c

Incr

easi

ng

mu

ltip

licit

yIncreasing pT

Project || > 2onto

!!! !!!


Associated yield grows with increasing multiplicity

Associated yield: correlated multiplicity per particle

Zero Yield At Minimum (ZYAM) • Data- PYTHIA8

N>1102.0<||<4.81GeV/c<pT<2GeV/c

Minimum of R

2.0<||<4.8

Quantifying the associated yield


Like-Sign vs. Unlike-Sign Pairs

No dependence on relative charge sign


Systematic Uncertainties, Checks

• Statistical uncertainty negligibly small

• However, the signal is subtle and unexpected

• Estimate systematic uncertainties

• Is there a way to fake the signal qualitatively?


Systematic Uncertainties

Analysis code

Reconstruction

Trigger

Detector

CMS Event

Collision

Physics

+ pile-up, beam backgrounds

+ detector noise, acceptance, efficiency

+ trigger efficiency, bias

+ efficiency, fakes

+ bugs?

Test the complete chain with data-driven checks!


Analysis Code

Independent codeSame definition of R

Same input file (skim)

Ridge is seen with three independent analysis codes

Standard analysis

Control analysis I

Control analysis II

Independent codeDifferent definition of R

Different input file (skim)

N>1101<pT<3GeV/c||>2


Reconstruction Code

“HighPurity” tracksPixel + Silicon Strip tracker

(Largely) independent codeIndependent detectors

Also: variation of tracking +vertexing parameters

Pixel-only tracks3 hits in pixel detector

N>1101<pT<3GeV/c


Trigger bias

Ridge is seen using min bias trigger + offline selection

Min bias trigger

Min-bias trigger vs. high mult trigger HLT 70 vs. HLT 85 for N > 110

No trigger bias seen from comparison of

trigger paths

N>1101<pT<2 GeV/c||>2

1<pT<2GeV/c||>2


Detector: -symmetry

Ridge is invariant under rotation

Ridge is not caused by rare events with large #

of pairs

Constrain one track to one -octant

• Signal• Background

Pair multiplicity distribution

for ||>2 and ||<1

N>1101<pT<2 GeV/c||>2

N>1101<pT<2 GeV/c||>2


Detector: uniformity in

Ridge region shows no structure in vs 2


Event Backgrounds

Ridge region shows no sensitivity to beam backgroundNote: Analysis is done on HighPurity tracks

Enrich the sample with beam-gas and beam-scraping events

Reject beam background by vetoon fraction of low quality tracks

Standard event selection

Increased beam scraping eventsIncreased beam halo

N>1101<pT<2GeV/c||>2

N>1101<pT<2GeV/c||>2No eff correction

CMS preliminary

CMS preliminary


Pileup collision events

No background or noise effects seen in cross-collision correlations

Correlate tracks from high multiplicity vertex with tracksfrom different collision (vertex) in same bunch crossing

N>110; 1 GeV/c<pT<3 GeV/c

2.0<||<4.8


31

Fra

ctio

n o

f pile

up

Pileup of collision events

Compare different run periods (fraction of pileup

varies by x4-5)

Change in pileup fraction by factor 2-4

has almost no effect on ridge signal

Compare different vertex regions

(fraction of pile-up ~ dN/dvtxz)

CMS preliminary

1<pT<2GeV/c||>2

CMS preliminary

CMS preliminary


Event Pileup

Pileup effects are suppressed due to excellent resolution

Track counting done with dz, dxy of O(100m)

Single-event track dz distribution

Track longitudinal and transverse impact

parameter (pT > 0.4 GeV/c)


Track-Photon Correlations

CMS preliminary

Note: photons reconstructed using “particle flow” event reconstruction technique

Use ECAL “photon” signalMostly single photons from 0’s

No efficiency, and pT, smearing corrections

N>1101<pT<3GeV/c||>2

N>1101<pT<3GeV/c


Photon-Photon Correlations

Use ECAL “photon” signalMostly single photons from 0’s

No efficiency, and pT, smearing corrections

Qualitative confirmationIndependent detector, independent reconstruction

CMS preliminary

N>1101<pT<3GeV/c||>2

N>1101<pT<3GeV/c


Systematic Uncertainties

Analysis code

Reconstruction

Trigger

Detector

CMS Event

Collision

Each step tested with data-based checks

No indication of effect that would fake ridge signal

Conservative estimates of uncertainties on ridge associated yield


Summary

• Study of short-range and long-range angular

correlations in pp collisions with CMS at LHC

• Observation of long-range, near-side correlations in

high multiplicity events

– Signal grows with event multiplicity

– Effect is maximal in the 1 < pT < 3 GeV/c range

– Not observed in low multiplicity events

– Not observed in MC generators

• This is a subtle effect in a complex environment –

careful work is needed to establish physical origin


Reconstructed high multiplicity pp event

Documents

The CMS detector Two-particle correlation functions Results in m inimum bias collisions