23
Pileup, substructure and other thoughts from CMS Nhan Tran Fermi National Accelerator Laboratory April 25th, 2013 Terascale Workshop on Substructure

Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

Embed Size (px)

Citation preview

Page 1: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

Pileup, substructure and other thoughts from CMS

Nhan TranFermi National Accelerator Laboratory

April 25th, 2013Terascale Workshop on Substructure

Page 2: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

outline

• Initially, discuss open question about the topic of substructure and its utility in mitigating pileup

• Last week, CMS had a substructure workshop where we also started thinking about future issues, post shutdown

• So, I will start by talking about pileup and then touch on a few other subjects • Should we be worried about this? Is it something only

CMS should worry about? Is it something experimentalists should worry about?

• Thinking about jet substructure applications beyond the classic boosted heavy object case

• N.B. this has a personal twist, not representative of all of CMS

2

Page 3: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

particle flow

3

CMS employs a particle flow algorithm translating detector measurements into physics objects

• Algorithm uses information from all subsystems, returns a list of muons, electrons, photons, and neutral and charged hadrons

• List is used as the building blocks for final physics objects in analyses:• Jets and b-jets, taus, and missing transverse energy (MET)

Page 4: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

jet clustering and corrections

4

CMS Collaboration

L1PU, MC

L3Absolute

Jets and Jet Energy Corrections

ALGORITHMS:

❖ Anti-kT 0.5 applied to all Particle Flow candidates (PFJets)

• Remove charged candidates not compatible with primary vertex (PFCHSJets)

CORRECTIONS (FACTORIZED APPROACH):

2

L1PU, data

L2Relative

L2L3Res

L5Flavor

MANDATORY FOR DATA AND MC

RESIDUALSMANDATORY FOR DATA

OPTIONAL

or

CMS standard are jets clustered with the Anti-kT (AK) algorithm with an R parameter R = 0.5 built from particle flow inputs (PF).

In addition, we are able to subtract charged hadrons from jets which do not come from the primary vertex (PFCHS Jets)

L1 corrects for pileup using FastJet ρ×A correction L2/L3 jet response versus pT, η

Dedicated corrections for AK5 jets and AK7 jets

Page 5: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

typical pileup contribution

5

2012 JEC status Approval Plots Conclusions Jet Corrections

Pile-Up Corrections

Both NPV-based and Fastjet-⇢-based corrections are in good agreement

Remaining Data/MC di↵erences accounted for with separate pile-up correctionsData/MC di↵erences mitigated by reweighing pile-up Poisson mean in MC to data.Poisson mean determined from measured luminosity and Minimum bias cross section.

Number of primary vertices0 5 10 15 20 25 30 35

pT p

ile-u

p (G

eV)

0

5

10

15

20

25

30

= 8 TeVs-1CMS preliminary, L = 1.6 fb

| < 0.5η|

R=0.5 PFlowTAnti-k

Average Offset (DATA)Average Offset (MC)Jet Area (DATA)Jet Area (MC)

Number of primary vertices0 5 10 15 20 25 30 35 40 45

a.u.

0.02

0.04

0.06

0.08

0.1 430 GeV≥ T

Jet p = 12.41〉

PV,dataN〈

= 12.44〉PV,MC

N〈

= 8 TeVs-1CMS preliminary, L = 1.6 fb

Ricardo Eusebi, on behalf of the JEC group: JEC 4 / 12CMS Collaboration

η-5 -4 -3 -2 -1 0 1 2 3 4 5

(Offs

et),

GeV

Tp

0

5

10

15

20

25

=5PVN=10PVN=15PVN=20PVN

PFJetsDataSimulation

CMS Preliminary = 8 TeVs

New Level 1 Corrections

❖ New ‘precision scaling’ in bins of (η, NPV)

3

2012 JEC status Approval Plots Conclusions Jet Corrections

Pile-Up Corrections

Pile-up measured with Zero Bias data and MC, then calibrated to QCD MC o↵set.Random cone method allows to separate contribution per subdetectorMost charged hadrons can be associated to pile-up vertices and removed

Part that can be removed is labeled ”charge Hadrons”Part that remains as PU after charge hadron substraction is labeled ”charged pile-up”

η-5 -4 -3 -2 -1 0 1 2 3 4 5

, GeV

〉 T,

offs

etp〈

0

0.2

0.4

0.6

0.8

1

1.2

1.4photonsem depositsneutral hadronshadronic depositscharged pile-upcharged hadrons

PVOffset/N

CMS preliminary = 8 TeVs

The CMS collaboration JEC 3 / 12

A few plots to give you a scale of the pileup on average

Page 6: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

pileup jets

6

Page 7: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

identifying jets from pileup

• Pileup jets are several lower energy jets not originating from the primary vertex which are top of each other

• Particular jet substructure variables are found to separate PU jets from real jets -- similar philosophy to quark/gluon separation• β = fraction of jet pT from charged tracks

coming from primary vertex (in tracking volume only)

• Jet width, shapes• Charged track multiplicity

7

“real”

“pileup”

12/06/12 5

Critical Variables (Jet id MVA/cuts)● Jet Shapes

● Multiplicities

● # of charged candidates (|η| < 2.75)

● # of neutral candidates

● Vertex compatabilitiy (|η| < 2.75)

● β = Σ pT(tracks |Δz| < 0.2 to PV)/total p

T

● β*= Σ pT(tracks |Δz| < 0.2 to other PV)/total p

T

11

The most powerful handle In the region covered by tracker region, the most powerful discriminant

are tracking information. Well summarized by the beta variable

Extremely good discrimination up to |h| = 2.5

Residual information up to |h|=2.75

β=∑Δ z (track , v0 )<0.2cm

pTcand

∑ pTcand

20 < pT < 30

TK HEin

Z → µµ

Page 8: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

sorry

8

No public plots yet, this is what it would look like...

Validation is done with Z+jets event balancing to isolate pileup jets

Find hard recoil back-to-back with Z and classify other jets as pileup

Warning: does not include quark/gluon differences

pileup jet, quark/gluon jet

β

N constituents frac ΔR 0.1

RMS

Page 9: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

identifying jets from pileup

• example, technique applied in Hττ analysis

• very little contamination from pileup jets in 0-jet category, even for very low jet pT

• reduces experimental jet veto uncertainties

9

threshold [GeV]T

Jet p20 40 60 80 100

Fra

ctio

n of

0-je

t eve

nts

0.5

0.6

0.7

0.8

0.9

1

# vtx all# vtx [0,8]# vtx [9,12]# vtx [13,16]# vtx [17,20]# vtx [20,+]

threshold [GeV]T

Jet p20 40 60 80 100

Rat

io0.98

1

1.02

5.1/fb 8 TeV, CMS Preliminary

Page 10: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

post LS1

10

Potential performance Number

of bunches

Ib LHC

FT[1e11]

beta*X beta*sep

Xangle

Emit LHC [um]

Peak Lumi [cm-2s-1] ~Pile-up

Int. Lumi per year

[fb-1]

25 ns 2760 1.15 55/43/189 3.75 9.2e33 21 ~24 25 ns

low emit 2320 1.15 45/43/149 1.9 1.5e34 42 ~40

50 ns 1380 1.65 42/43/136 2.5 1.6e34 level to 0.9e34

74 level to

40 ~45*

50 ns low emit 1260 1.6 38/43/115 1.6

2.2e34 level to 0.9e34

109 level to

40 ~45*

• 6.5 TeV • 1.1 ns bunch length • 150 days proton physics, HF = 0.2 • 85 mb visible cross-section • * different operational model – caveat - unproven All numbers approximate 40

Mike Lamont

could have essentially double pileup vertices

post LS2 could be > 100 PU

Page 11: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

post LS1

11

Potential performance Number

of bunches

Ib LHC

FT[1e11]

beta*X beta*sep

Xangle

Emit LHC [um]

Peak Lumi [cm-2s-1] ~Pile-up

Int. Lumi per year

[fb-1]

25 ns 2760 1.15 55/43/189 3.75 9.2e33 21 ~24 25 ns

low emit 2320 1.15 45/43/149 1.9 1.5e34 42 ~40

50 ns 1380 1.65 42/43/136 2.5 1.6e34 level to 0.9e34

74 level to

40 ~45*

50 ns low emit 1260 1.6 38/43/115 1.6

2.2e34 level to 0.9e34

109 level to

40 ~45*

• 6.5 TeV • 1.1 ns bunch length • 150 days proton physics, HF = 0.2 • 85 mb visible cross-section • * different operational model – caveat - unproven All numbers approximate 40

Mike Lamont

could have essentially double pileup vertices

post LS2 could be > 100 PU

This does not even take into account out-of-time pileup which CMS will be more affected by at 25 ns.

Initial reaction, what’s the future of 25-45 GeV jets post LS1? in the forward region? post LS2?

Page 12: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

detector considerations

12

• Pileup mitigation• Depth segmentation in the HCAL,

different depths in the hadronic calorimeter can be used to identify pileup contributions• Currently in the HE only but by LS2

for the whole HCAL • Timing information in HCAL (post LS2) -

can help in identifying out-of-time pileup

• Really high pT jets - more prevalent at √s = 13 TeV with more boosted objects• When does tracking start to fail? At which momentum do the

tracks become too close together• The outer region of the HCAL is not being used currently (HO),

how much will this improve the performance of really high pT jets?

Page 13: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

pileup mitigation with grooming

13

• Grooming algorithms are used to clean up soft QCD and pileup contributions to the jet

• Can we use grooming algorithms to reduce the affect of pileup on our “standard” AK5 jets?

• Typically grooming algorithms reduce the jet area, reducing the size of the “L1” pileup correction

• Which grooming algorithm should we use? • Dedicated studies are needed, varying

grooming parameters• Pruning is standard for searches with

substructure but also the most “invasive” to jets, maybe trimming is a better choice

• Are there situations where a CMS standard jet in physics analyses was a groomed jet is better than a regular one?• e.g. Jet vetoes?

RECOd

-2 0 2

>G

ENun

groo

med

/pT

REC

Oun

groo

med

<pT

>G

ENgr

oom

ed/p

TR

ECO

groo

med

<pT

doub

le ra

tio,

0.8

0.9

1

1.1

Ungroomed AK7Trimmed AK7Filtered AK7Pruned AK7

= 7 TeV, AK7 W+jetss at -1CMS Simulation, L = 5fb

RECOη

-20

2

>GENungroomed/pT

RECOungroomed<pT

>GENgroomed/pT

RECOgroomed<pT

double ratio,

0.8

0.9 1

1.1

1.2

ungroomed

trimm

edfilteredpruned

CM

S Preliminary 2011

-1 = 7 TeV, L = 5.02 fb

s

CMS Collaborationjetη

-4 -2 0 2 4

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFTAnti-k=30 GeV

Tp

= 8 TeVs-1CMS preliminary, L = 11 fb

jetη

-4 -2 0 2 4

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFTAnti-k=100 GeV

Tp

= 8 TeVs-1CMS preliminary, L = 11 fb

jetη

-4 -2 0 2 4

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFchsTAnti-kE=1000 GeV

= 8 TeVs-1CMS preliminary, L = 11 fb

(GeV)T

p20 100 200 10002000

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFTAnti-k|=0

jetη|

= 8 TeVs-1CMS preliminary, L = 11 fb

(GeV)T

p20 100 200 10002000

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFTAnti-k|=2.0

jetη|

= 8 TeVs-1CMS preliminary, L = 11 fb

(GeV)T

p20 100 200 10002000

JEC

unc

erta

inty

[%]

012

34

56789

10Total uncertaintyAbsolute scaleRelative scaleExtrapolationPile-up, NPV=14Jet flavorTime stability

R=0.5 PFTAnti-k|=2.7

jetη|

= 8 TeVs-1CMS preliminary, L = 11 fb

JEC Uncertainty (PFJets)

8

vs p

Tvs

η

Page 14: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

scanning in grooming parameters

14

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈

20

40

60

80

100

120

140

160 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

LCW jets with R=1.0tanti-k| < 0.8η < 300 GeV, |

Tjet p≤200

No jet grooming =0.3sub=0.01, Rcutf=0.3sub=0.03, Rcutf =0.3sub=0.05, Rcutf=0.2sub=0.01, Rcutf =0.2sub=0.03, Rcutf=0.2sub=0.05, Rcutf

(a) Trimmed anti-kt: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

6080

100120140160180200220240260280 ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, LCW jets with R=1.0tanti-k

| < 0.8η < 800 GeV, |Tjet p≤600

No jet grooming =0.3sub=0.01, Rcutf=0.3sub=0.03, Rcutf =0.3sub=0.05, Rcutf=0.2sub=0.01, Rcutf =0.2sub=0.03, Rcutf=0.2sub=0.05, Rcutf

(b) Trimmed anti-kt: 600 pjetT < 800 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈

60

80

100

120

140ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, LCW jets with R=1.0tanti-k

| < 0.8η < 300 GeV, |Tjet p≤200

No jet grooming =0.05cut

=0.10, zcutR=0.10

cut=0.10, zcutR =0.05

cut=0.20, zcutR

=0.10cut

=0.20, zcutR =0.05cut

=0.30, zcutR=0.10

cut=0.30, zcutR

(c) Pruned anti-kt: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

100

120

140

160

180

200

220

240 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

LCW jets with R=1.0tanti-k| < 0.8η < 800 GeV, |

Tjet p≤600

No jet grooming =0.05cut

=0.10, zcutR=0.10

cut=0.10, zcutR =0.05

cut=0.20, zcutR

=0.10cut

=0.20, zcutR =0.05cut

=0.30, zcutR=0.10

cut=0.30, zcutR

(d) Pruned anti-kt: 600 pjetT < 800 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈

0

20

40

60

80

100

120

140

160 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

C/A LCW jets with R=1.2| < 0.8η < 300 GeV, |

Tjet p≤200

No jet grooming=0.67

fracµ

=0.33fracµ

=0.20fracµ

(e) Filtered C/A: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

50

100

150

200

250

300ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, C/A LCW jets with R=1.2

| < 0.8η < 800 GeV, |Tjet p≤600

No jet grooming=0.67

fracµ

=0.33fracµ

=0.20fracµ

(f) Filtered C/A: 600 pjetT < 800 GeV

Figure 1: Evolution of the mean jet mass, hmjeti, for jets in the central region |⌘| < 0.8 as a function ofthe reconstructed vertex multiplicity, NPV for leading jets in the range 200 pjet

T < 300 GeV (left) andthe range 600 pjet

T < 800 GeV (right). (a)-(b) show trimmed anti-kt jets with R = 1.0, (c)-(d) showpruned anti-kt jets with R = 1.0, and (e)-(f) show split and filtered C/A jets with R = 1.2. The error barsindicate the statistical uncertainty on the mean value in each bin.

6

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈

20

40

60

80

100

120

140

160 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

LCW jets with R=1.0tanti-k| < 0.8η < 300 GeV, |

Tjet p≤200

No jet grooming =0.3sub=0.01, Rcutf=0.3sub=0.03, Rcutf =0.3sub=0.05, Rcutf=0.2sub=0.01, Rcutf =0.2sub=0.03, Rcutf=0.2sub=0.05, Rcutf

(a) Trimmed anti-kt: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

6080

100120140160180200220240260280 ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, LCW jets with R=1.0tanti-k

| < 0.8η < 800 GeV, |Tjet p≤600

No jet grooming =0.3sub=0.01, Rcutf=0.3sub=0.03, Rcutf =0.3sub=0.05, Rcutf=0.2sub=0.01, Rcutf =0.2sub=0.03, Rcutf=0.2sub=0.05, Rcutf

(b) Trimmed anti-kt: 600 pjetT < 800 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈60

80

100

120

140ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, LCW jets with R=1.0tanti-k

| < 0.8η < 300 GeV, |Tjet p≤200

No jet grooming =0.05cut

=0.10, zcutR=0.10

cut=0.10, zcutR =0.05

cut=0.20, zcutR

=0.10cut

=0.20, zcutR =0.05cut

=0.30, zcutR=0.10

cut=0.30, zcutR

(c) Pruned anti-kt: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

100

120

140

160

180

200

220

240 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

LCW jets with R=1.0tanti-k| < 0.8η < 800 GeV, |

Tjet p≤600

No jet grooming =0.05cut

=0.10, zcutR=0.10

cut=0.10, zcutR =0.05

cut=0.20, zcutR

=0.10cut

=0.20, zcutR =0.05cut

=0.30, zcutR=0.10

cut=0.30, zcutR

(d) Pruned anti-kt: 600 pjetT < 800 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

jet

m〈

0

20

40

60

80

100

120

140

160 ATLAS Preliminary-1 Ldt = 1 fb∫Data 2011,

C/A LCW jets with R=1.2| < 0.8η < 300 GeV, |

Tjet p≤200

No jet grooming=0.67

fracµ

=0.33fracµ

=0.20fracµ

(e) Filtered C/A: 200 pjetT < 300 GeV

)PV

Reconstructed vertex multiplicity (N0 2 4 6 8 10 12 14

[GeV

]〉

1jet

m〈

50

100

150

200

250

300ATLAS Preliminary

-1 Ldt = 1 fb∫Data 2011, C/A LCW jets with R=1.2

| < 0.8η < 800 GeV, |Tjet p≤600

No jet grooming=0.67

fracµ

=0.33fracµ

=0.20fracµ

(f) Filtered C/A: 600 pjetT < 800 GeV

Figure 1: Evolution of the mean jet mass, hmjeti, for jets in the central region |⌘| < 0.8 as a function ofthe reconstructed vertex multiplicity, NPV for leading jets in the range 200 pjet

T < 300 GeV (left) andthe range 600 pjet

T < 800 GeV (right). (a)-(b) show trimmed anti-kt jets with R = 1.0, (c)-(d) showpruned anti-kt jets with R = 1.0, and (e)-(f) show split and filtered C/A jets with R = 1.2. The error barsindicate the statistical uncertainty on the mean value in each bin.

6

It would be great if CMS does something like this. Does it make an effect for smaller R jet? With area subtraction?

Page 15: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

inclusive measurements

15

R

observable

0.5

1.0

jet m

ass

[pru

ned]

jet m

ass

[trim

med

]

jet m

ass

[(mas

s-dro

p)/f

ilter

ed]

AK&CA

V+jets and dijets final statesAll measurements are unfolded5 fb-1 at 7 TeV, 2011

Dijet final state⦿ = unfolded, ◎ = detector level❖ = 36 pb-1 at 7 TeV, 2010✴ = 5 fb-1 at 7 TeV, 2011, grooming parameters varied

N.B. jet finding with anti-kT (AK) unless otherwise indicated, alternative algorithm: Cambridge-Aachen (CA)

jet m

ass

kT sp

littin

g sc

ale

and

N-su

bjet

tines

s,

τ 2/τ

1 and

τ 3/τ

2

⦿❖,CA

⦿❖,AK

⦿❖,CACA

⦿❖,CA

widt

h, e

ccen

tricit

y,

plan

ar fl

ow,

angu

larit

y

⦿❖,CA ⦿❖

⦿❖⦿❖

◎✴ ◎✴

◎✴,CA◎✴,CA

◎✴◎✴◎✴

◎✴◎✴ ◎✴◎✴

Page 16: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

inclusive measurements

16

R

observable

0.5

1.0

jet m

ass

[pru

ned]

jet m

ass

[trim

med

]

jet m

ass

[(mas

s-dro

p)/f

ilter

ed]

AK&CA

V+jets and dijets final statesAll measurements are unfolded5 fb-1 at 7 TeV, 2011

Dijet final state⦿ = unfolded, ◎ = detector level❖ = 36 pb-1 at 7 TeV, 2010✴ = 5 fb-1 at 7 TeV, 2011, grooming parameters varied

N.B. jet finding with anti-kT (AK) unless otherwise indicated, alternative algorithm: Cambridge-Aachen (CA)

jet m

ass

kT sp

littin

g sc

ale

and

N-su

bjet

tines

s,

τ 2/τ

1 and

τ 3/τ

2

⦿❖,CA

⦿❖,AK

⦿❖,CACA

⦿❖,CA

widt

h, e

ccen

tricit

y,

plan

ar fl

ow,

angu

larit

y

⦿❖,CA ⦿❖

⦿❖⦿❖

◎✴ ◎✴

◎✴,CA◎✴,CA

◎✴◎✴◎✴

◎✴◎✴ ◎✴◎✴

only one overlapping point ,CA12, but unfortunately here CMS did filtering only and ATLAS did split/filtering method

#facepalm

Would grooming help here?

Page 17: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

example: very high PU

• We can even glance into the far future with Snowmass studies for very high PU scenarios (Phase 2, 140 PU)

• Grooming more than just a improvement, rather it maybe become a necessity

• Snowmass setup:ttbar sample with 0 and 140 PU using the DELPHES detector

• Attempt to pick out the merged W’s for moderatelyboosted tops

• W peak is washed out without grooming

17

J. Dolen

Snowmass Energy Frontier Workshop - April 5, 2013

)2Jet mass (GeV/c0 20 40 60 80 100 120 140 160 180 200

Num

ber o

f Jet

s

0

50

100

150

200

250

300

350 0 PU - no grooming

140 PU - no grooming

140 PU - pruned

Jet Mass

21

• Select high pT leading jets (pT>300)

• W peak is visible in 0 PU TTbar jet mass distribution

- We need improved high pT statistics and larger jets to reconstruct top jets

• Peak washed out in 140 PU sample

• Pruning recovers the peak

More details:https://indico.bnl.gov/getFile.py/access?contribId=19&sessionId=10&resId=0&materialId=slides&confId=571

Page 18: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

high pT and high PU

• What happens at both high pT and high pileup?• Which tagging techniques work in the region we also

worry about detector resolution?

18

Comparison of taggers

But only for a limited range of masses !

0

0.1

0.2

0.3

10-6 10-4 0.01 0.1 1

10 100 1000

l/m

dm

/ dl

l = m2/(pt2 R2)

gluon jets: m [GeV], for pt = 3 TeV

Jets: C/A w

ith R=1. M

C: Pythia 6.4, D

W tune, parton-level (no M

PI), ggAgg, pt > 3 TeV

plain jet massTrimmerPrunerMDT

0

0.1

0.2

0.3

10-6 10-4 0.01 0.1 1

10 100 1000

l/m

dm

/ dl

l = m2/(pt2 R2)

quark jets: m [GeV], for pt = 3 TeV

Jets: C/A w

ith R=1. M

C: Pythia 6.4, D

W tune, parton-level (no M

PI), qqAqq, pt > 3 TeV

plain jet massTrimmer (zcut=0.05, Rsub=0.2)

Pruner (zcut=0.1)

MDT (ycut=0.09, µ=0.67)

Marzani et al.

Page 19: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

example: Q-jets, Q-event, telescoping

• Could Qjets be a good measure for identifying pileup jets?• Speed can be problem...

• Q-events, telescoping: • It should be particularly useful in high multiplicity

signatures (SUSY)• It could be useful in identifying VBF tag jets where

we have no tracking information and lots of pileup

19

Page 20: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

new observables

20

what observable would you like to see that you aren’t an author of?

(Jon)

• Q-jets, etc• jet charge• shower deconstruction• N-jettiness as a veto, algorithm?• Energy correlation functions• ...

for theorists, what kind of measurements would you like to see being done?

Page 21: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

MC tools

• Not optimal right now• Only Pythia6 is tuned by CMS, but does not describe jet substructure• What's the status/timescale of NLO ME+PS MCs?

• Sherp 2.0 (MEPS@NLO), POWHEG + MINLO, the GENEVA project• Still, we wouldn’t be able to use these for all samples

• We are moving to Pythia8 post LS1, no tune exists at the moment• CMS/ATLAS have unfolded jet mass distributions

• This is our chance to impact the parton shower tunes to get better modeling of jet substructure

• Request from theorists of a CMS paper on jet substructure • Compare data to different types calculations

• LO+PS, LO+Matching+PS, NLO+Matching+PS, alternate PS+UE+Hadronization+PU

• May systematically help understanding which calculations (don’t) work

• Compare generator level and detector level• Theorists want to know which information in their variables

survive the detector reconstruction21

Page 22: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

summary

• A lot of open questions about pileup post LS1• Experiments will start to answer some of these questions

soon, samples are becoming available to do detailed studies

• In the meantime, get input on what are most important to think about

• CMS has started to apply substructure ideas in identifying pileup jets, how much further can we take it?

• Thinking about other applications for substructure beyond “classic” heavy object tagging

22

Page 23: Pileup, substructure and other thoughts from CMSpages.uoregon.edu/soper/TeraJets2013/TeraJets2013Talks/tran.pdf · Pileup, substructure and other thoughts from CMS ... PU, MC L3 Absolute

CMS: Compact Muon Solenoid

23

3.8T Solenoid

ECAL!76k scintillating PbWO4 crystals HCAL!Scintillator/brass

Interleaved ~7k ch

•  Pixels (100x150 µm2) " ~ 1 m2 ~66M ch"• Si Strips (80-180 µm)" ~200 m2 ~9.6M ch!

Pixels'&'Tracker!

MUON'BARREL!250 Drift Tubes (DT) and 480 Resistive Plate Chambers (RPC)

473 Cathode Strip Chambers (CSC) 432 Resistive Plate Chambers (RPC)

MUON'ENDCAPS!

Total+weight+++++++++14000+t+Overall+diameter+++15+m+Overall+length+++++++28.7+m+

IRON'YOKE'

YBO YB1-2

YE1-

3

Preshower Si Strips ~16 m2

~137k ch

Foward Cal Steel + quartz Fibers 2~k ch