Using Object Correlations to Extract New Physics from the LHC

Rouven Essig, Michael Graesser, Eva Halkiadakis, Dmitry Hits,

Amit Lath, Pablo Mosteiro, Keith Rose, Steve Schnetzer,

Jessie Shelton, Sunil Somalwar, Scott Thomas

Rutgers HEX + HET

(Reconstructed) Objects

Leptons Photons Missing Transverse Energy Jets

Exotic Objects (New Long Lived Particles)

Displaced - Leptons, Photons, Jets Highly Ionizing Tracks Highly Ionizing to Minimum Ionizing Kinks Highly Ionizing Stopped Track Out of Time Decays Charge Exchange Tracks Charge Changing Tracks .....

(Reconstructed) Objects

Leptons Photons Missing Transverse Energy Jets

New Physics Searches – Hadron Collider

Strong Production Cross Section . Electroweak Decays . Background / Signal Suppressed

pp QQ) / (pp Jets)

Q Leptons, Photons, MET

(Reconstructed) Objects

Leptons Photons Missing Transverse Energy Jets

New Physics Searches – Hadron Collider

Strong Production Cross Section . Decay to Strong States . Background / Signal Severe

New Physics Search in Multi-jets :

pp QQ) / (pp Jets)

Q Jets

• Challenging - ....

• But Possible - New Physics May be Hidden in Jets … (c.f. Higgs at LEP)

• Techniques may be Useful for Multi-jets in Association . with Leptons, Photons, MET

New Physics Searches in Multi-Jets

j j j j j j

pp QQ

• Magnitude of Multi-Jet Backgrounds from High Order Processes . Difficult to Calculate a priori s

n

• High Multiplicity-Jet Backgrounds, Lower Rate - More Tractable

• Kinematic Features + Correlations

R-Parity Violation uud Yukawa

Kinematic Features + Correlations + Hadronic Top Guidance . (No b-tagging or mW resonance )

• Hiding New Physics in Multi-Jets

• SUSY gluino j j j

Signal : Pythia

6 Jet Background : ALPGEN Pythia

Hadronic Top Background : Pythia

Detector Simulator : PGS

Analysis : ChRoot +

Q = SU(3)C Adjoint Majorana Fermion

Semi-Leptonic Top Decays

pp tt l j j j j

Njet ¸ 4

Ne = 1

pT,4th jet ¸ 40 GeV

PT,e ¸ 20 GeV

| mjjj – mW | · 10 GeV

j j j Furthest R from Electron

pp W j j j j

GeV

GeV

PT,5th jet · 10 GeV

mjjj

150 pb-1

150 pb-1

Gluino300-E

ffjet6Effjet

Alp6j50-E

ffjet6Effjet

pp QQ j j j j j j

Njet ¸ 6

mQ = 420 GeV

PT, jets PT, jets

PT

,6th

jet

PT

,6th

jet

2 fb-1

Signal Hadronic Background

0.015 fb-1

Gluino300-E

ffjet6Effjet

Alp6j50-E

ffjet6Effjet

pp QQ j j j j j j

Njet ¸ 6

Cut

mQ = 420 GeV

Cut

PT, jets PT, jets

PT

,6th

jet

PT

,6th

jet

Cut :

pT, jets ¸ 700 GeV

pT, 6th jet ¸ 60 GeV

2 fb-1

Signal Hadronic Background

Signal Efficiency ' 0.25

0.015 fb-1

pp QQ j j j j j j

Two Three-Body Resonances mjjj = mjjj

| i=16 pT,i,jet | · 60 GeV

Choose Pair of Jet triplets with Smallest | mjjj – mjjj | · 60 GeV

mjjj mjjj

Signal Hadronic Background

• Cuts Necessarily Shape Background

• Magnitude of Background Uncertain - . Use Signal + Background Distributions as Templates

2 fb-1 0.015 fb-1

pp QQ j j j j j j

Scaled Cuts pT, jets ¸ 1.7 mQ . pT, 6th jet · 0.15 mQ . | i=1

6 pT,i,jet | · 0.15 mQ

| mjjj – mjjj | · 0.15 mQ

MQ (GeV) S/B S/ B1/2 / fb-1 Signal Events / fb-1

420 1/29 17 9000

660 1/22 5 600

890 1/17 2 70

Note: Magnitude of Background Uncertain

pp QQ j j j j j j

| i=16 pT,i,jet | · 60 GeV

mjjj

mjjj

Signal Hadronic Background

Tail much larger than Jet Resolution - . Mismatching of Jet Triplets - . Combinatoric Background Within Signal

Improve Contrast Between Signal and Background

Choose Pair of Jet triplets with Smallest | mjjj – mjjj | · 60 GeV

pp QQ j j j j j j

Reduce Combinatoric Background from Radiated Jets with Cut on 7th Jet

pT,jets ¸ 500 GeV pT, 6th jet ¸ 30 GeV pT,

7th jet · 20 GeV | i=16

pT,jeti | · 50 GeV

MQ = 290 GeV

Signal Hadronic Background

mjjj mjjj

MQ (GeV) S/B S/ B1/2 / fb-1 Signal Events / fb-1

290 1/110 15 26850

• Improves Contrast AND . S/B by Factor 2

0.3 fb-1 0.09 pb-1

pp QQ j j j j j j

Kinematic Correlation – Separate Signal from Combinatoric Background

mQ = 420 GeV

Signal Signal

pT,jets

mjjj

mjjj

| i=13 pT,jeti

|

• Two Distributions of Events

• Three Body Resonance Jet Resolution Apparent on Horizontal Branch

• Most of the “Best” Pairs of Jet Triplets with small |mjjj-mjjj| . are NOT the Correct Pairing (Random Choice of Triplet » Same Distribution)

pp QQ j j j j j j

Remove Combinatoric Background with Cuts

Signal Signal

mjjj

mjjj

• Horizontal Branch - Region of High Signal to Combinatoric . Background Contrast

- Increase Signal Efficiency by Including All 20 Jet Triplets

- Remove | mjjj – mjjj | Cut ( | i=16 pT,jeti

| Cut )

pT,jets | i=13 pT,jeti

|

pp QQ j j j j j j

Hadronic Top Background Hadronic Top Background

Cuts

Increase Signal Efficiency

Hadronic Background Hadronic Background

mjjj

mjjj

PT,jets | i=13 pT,jeti

|

mjjj

mjjj

pp QQ j j j j j j

Signal Hadronic Background

Hadronic Top Background

mjjj

mjjj

0.3 fb-1

2 fb-1 0.015 fb-1

pp QQ j j j j j j

Signal Hadronic Background

mjjj

MQ (GeV) S/B S/ B1/2 / fb-1 Signal Jet Triplets / fb-1

490 1/15 17 4700

• Improves Constrast and S/B by 2.5 but Lowers Signal Efficiency by 2.8

2 fb-1 0.015 fb-1

pp QQ j j j j j j

Using Correlations –

Possible to Separate pp QQ j j j j j j from Background

Optimize Cuts

Study Template Fitting to Signal + Background

Validate Background Templates and Magnitude with Data

Possible Reach (Preliminary)

mQ » 600 GeV with 1 fb-1

» 800 GeV with 10 fb-1

Develop Kinematic Correlations for Other Multi-Jet Signatures

Extend to Signatures with Other Objects - Correlation Templates

Kinematic Correlations

Decay Trees

Invariant Kinematic Correlations = f( mijk…2) (Unpolarized,

. Spins Unobserved, . T-Invariance ) mijk…

2 = f(mij2)

Correlations in Generalized Dalitz Space mij2 i,j = All Pairs of Objects

Invariant Kinematic Correlations

• New Physics Sample . Nl = 2-6 . Njet ¸ 2 . Backgrounds Unimportant

m12

m1

3

Nl ¸ 3

• Enhanced S/BCombinatoric Constrast

DS < DB (D=1 Histrogram DS = DB)

• Two Body Resonant Tree

• Correlation m122 + m13

2 = Constant

(Some of) The Two Body Resonances . Arise from Three Body Decay

6 fb-1

Invariant Kinematic Correlations

m122

m1

32

2

2

Dalitz (symmetrized)

• Uniform |M|2 on Two Body Resonances

Consistent with Intermediate Scalar

• Edges

Three Body Tree (Only feature in . D=1 Histogram)

Coincide with Endpoint of Resonant . Two Body Correlation

- Arise from “Missing” Lepton

- Anything Else Contributing to . Edge is massless

• Relative Density of Two Body Resonances, . Resonant Two Body Correlation, and . Edge Contrast

- Br ( lll) / Br( llX)

m122

m1

32

Invariant Kinematic Correlations

Three Body Resonant Trees

310 GeV · mlll · 330 GeV

• Further Enhances . S/BCombinatoric Contrast

m1

23

Kinematic Correlations

m23

m1

4

• No Kinematic Correlation

- Consistent with Arising from . Different Parent Particle

• m14 = m and m23 m or

m14 m and m23 = m

- Single Resonant Two Body Decay

• m14 = m23 = m

- If Density at Intersection \geq 2 . Some Events with (At least) . Two Resonant Two Body Decays

- Density at Intersection gives Fraction

Nl ¸ 4

30 fb-1

Mjlll

Invariant Kinematic Correlations

Nj ¸ 2M

jlll

Mjlll

• Contrast S/BCombinatoric Improved

• Two Four Body Resonant Decays

• Density in Bands and Intersections Give

Br( 1 jlll) / Br( 2 jlll)

• Correlation Could Indicate Resonant . Five Body Decay jjlll

Object Correlations to Extract Low Order Trees (OCELOT)

Kinematic Correlations Can Enhance Contrast DS < DB

Menu of Correlations for Low Order Trees

- Develop Templates to Search for Correlations

Extend to Trees with MET

(Generalization to D > 1 of Edges and Endpoints)

Correlations Allow Direct Measurements

Implement Correlations in Fitting Procedure to Decay Trees . (Fitting to D=1 Counts Misses Many Correlations)

New Physics at the LHC

Resonant Multi-Jets ….. Resonant Multi-Leptons

Very Hard Very Easy