model-independent approaches to missing energy at the...

Model-independent approaches to missing energy at the LHC

Johan Alwall, SLAC

NTU, Taipei, Taiwan, 16 Sep 2009

Based on:arXiv:0810.3921 (PRD 79, 075020)arXiv:0809.3264 (PRD 79, 015005)

Johan Alwall - Model-independent approaches at the LHC 2

● The Standard Model – one of the most successful theories in the history of physics

Introduction: The Standard Model




● Describes the particle content and interactions of the microscopical world with no major discrepancies with experiment




● Describes the particle content and interactions of the microscopical world with no major discrepancies with experiment

● Yet undiscovered: The Higgs particle, breaks the Electroweak symmetry and gives mass to all particles



Problems with the Standard Model● Quadratic quantum corrections to the mass of

the Higgs particle → hierarchy problem

t

t

H H ΔMH

2 ~ M2new physics

~100 GeV MPl

~1019 GeV




the Higgs particle → hierarchy problem ● Dark matter observations in the sky

Luminous (X-ray-emitting) gas(stopped by collision)

Dark matter (deduced by graviational lensing)(unaffected by collision)

MACS J0025.4-1222 cluster collision (Bradac et al)




the Higgs particle → hierarchy problem ● Dark matter observations in the sky

● Grand unification



Solutions to hierarchy problem● New weakly interacting particles and

symmetries that cancel the “bad” loops

● Composite Higgs (new strong interactions, Technicolor)

● Removing the hierarchy by strengthening gravity (extra spacial dimensions)

t

t

H H ~ M2t - M2

t

t

H H

~

+ ~

~1 TeV 175 GeV



Solutions to dark matter● New particles + symmetry that ensures stability

of lightest new particle (WIMP)– Simplest version: Internal parity under which SM

particles are even, new particles odd

Solutions to unification● New particles between

weak scale and GUT scale – modifies coupling evolution


New Physics at the LHC

“Favourite” model: Supersymmetry– Symmetry between fermions and bosons –

automatically stabilizes the Higgs mass to all orders

– SM particle ↔ “superpartner” – same quantum numbers and SM couplings, different spin and massquark ↔ squark gluon ↔ gluino top ↔ stop (etc.)

– Extra quantum number carried by partners (R-parity)● Stable lightest superpartner (LSP) → Dark matter candidate● SUSY particles always pair produced

– Allows to solve all three problems within one framework!



Supersymmetry not yet found (and has problems) – Need to consider alternatives

– Little Higgs

– Randall-Sundrum

– ADD (large extra dimensions)

– …

➔ To solve Hierarchy problem and dark matter, need similar features as SUSY– But can have very different mass spectra and

coupling structures!


So why the Large Hadron Collider?



So why the Large Hadron Collider?

● Particle masses around 100 GeV- 1 TeV – within reach for the 14 TeV LHC

● Some particles charged under QCD – easy to produce at a hadron collider

● We are (quite) confident that something new will be discovered at the LHC!



“SUSY-like” MET signatures● At the LHC, mainly particles charged under

QCD (quark and gluon partners) will be directly produced

● Decay through “cascades” emitting quarks and leptons until reach dark matter particle

● Signature: High-E hadron jets, leptons and missingtransverse momentum (MET)

Typical process in

SUSY!


Difficulties with MET signatures

● No visible resonances – no simple features above Standard Model backgrounds

● Event reconstruction not possible

● Complicated to measure jets and missing transverse momentum

● Difficult determining overall mass scale – only mass differences readily observable

● Extra hard jets from QCD radiation, besides the decay products – difficulties in combinatorics


Model-independent approaches to MET+jets/leptons signals

Many models/ideas for physics beyond the Standard Model at the LHC giving signatures with

leptons, jets and missing energy● Supersymmetry: over 120 free parameters /

many mediation scenarios (SUGRA, GMSB, AMSB, Gaugino mediation, Mirage mediation, non-standard scenarios, ...)

● Little Higgs: dozens of models● Extra-dimensional scenarios (RS, UED)




leptons, jets and missing energy● Differences between models often subtle –

difficult to distinguish in early data● Many models (esp. SUSY) have many

parameters – most unmeasurable at LHC● Constrained versions of models introduce

relations between observables not necessarily seen in data




leptons, jets and missing energy

How to analyze and present excesses – Without model bias– Possible to compare to any model

?


Present approaches

● Most common: Exclusions/analysis in model space of minimal model (mSUGRA/mGMSB/mAMSB) with few (~4) parameters

● Benchmark analyses (e.g. Snowmass points)● Signature-based exclusions (cross section

limits given some “standard” set of cuts)● In case of excesses: Scans of SUSY space

(~20 param) using high-level kinematical information


Present approaches

● Most common: Exclusions/analysis in model space of minimal model (mSUGRA/mGMSB/mAMSB) with few (~4) parameters

● Problems:– Fixed relations between parameters, e.g.

mg:m

W:m

B ~ 6:2:1

– Fixed decays and branching ratios

➔ Not all parameter space covered by analysis

~LSP

~ ~


Present approaches

mSUGRA mass ratiog → qqB

g → qqW+B

~ _~

~ _ ~mSUGRA exclusion (D0)

Example: mSUGRA/mGMSB assumes fixed ratio between gluon partner and photon partner➔ No Tevatron study allowing free mass ratio

J.A., Le, Lisanti, Wacker [arXiv:0803.0019]


Present approaches

● Benchmark analyses (e.g. Snowmass points)– Only relevant to assess power of methods

● Signature-based exclusions (cross section limits given some “standard” set of cuts)– Restricted in scope, reduced power

● In case of excesses: Scans of SUSY space (~20 param) using high-level kinematical information and rate information– Assumes SUSY, need high-statistics data


Presenting experimental data

Twofold problem:● How to analyze/report limits on cross sections

in a way that can be compared (by theorists) with any model?

● How to analyze/report/characterize stable excesses over background in a way that can be compared to any model and give relevant information on the underlying physics without bias to a certain model?


J.A., Schuster, Toro [arXiv:0810.3921]


Presenting experimental data

Twofold problem:● How to analyze/report limits on cross sections

in a way that can be compared (by theorists) with any model?

● How to analyze/report/characterize stable excesses over background in a way that can be compared to any model and give relevant information on the underlying physics without bias to a certain model?


J.A., Schuster, Toro [arXiv:0810.3921]

This talk

In backup


Model-independent characterization of excesses

If excesses in MET+jets/leptons are found at the LHC (signatures in leptons, jets and missing E

T),

the first questions* we want answered are:● Consistent with pair-production of particles with

cascade decays to massive dark matter?● Which colored particles dominate production?● Which decay chains are present?● Cross sections, mass scales?

Precision physics will come later!

*First ~500 pb-1(?)


Model-independent characterization of excesses

How do this in a well-defined matter without introducing model bias in analysis?

● Small set of “simplified models” addressing the specific questions we want answered

● Well-defined method to compare any theoretical model to the simplified models

➔ The simplified models can act both as a detector-independent representation of data andas a basis for information about possible model space consistent with the data


Simplified models

● Each simplified model: 3-4 mass parameters and 3-4 cross sections/branching ratios– All parameters directly measured in data

● One set of models to determine decay chains involving leptons

● One set of models to determine b quark content● For each case, one model with quark partners as

start of decay chain, one with gluon partners➔ Surprisingly good description of “data” generated

also with quite complicated models


Simplified models for lepton content

Lep(Q)

*on- or off-shell

Lep(G)

g

g

g

g*on- or off-shell


Simplified models for b quark content

Btag(Q)

Btag(G)


Fitting examples

Two examples presented in our paper:● One “simple” example, where structure of

underlying model mirrors the simplified models● One “complicated” example, where underlying

model has complicated structure, including multi-level decays, squeezed spectra, competing decays

➔ Simplified models give good fit also to data from complicated model!


Fitting examples

Simple example

Q

Johan Alwall - First Characterization of Excesses at the LHC 32

Fitting examples

Dilepton invariant massHT (effective mass)

Examples of plots used for mass determination:

Lep(G) (MG=600 GeV)Lep(Q) (MQ=500 GeV)



Fitting examples

Number of leptons in lepton-incl.signal region

OSO

F

OSSF

Z cand

SSOF

SSSF

Examples of plots used for lepton BR fitting:




Number of b jets above 75 GeV

Fitting examples

In this example: Lep(G)/Btag(G) “perfect fit” – can draw strong conclusions about underlying physics

Number of jets above 75 GeV

Examples of plots used for b content fitting and jet structure

Lep(G) (MG=600 GeV)Lep(Q) (MQ=500 GeV)Btag(G) (MG=600 GeV)

Johan Alwall - Characterization of Early Excesses 35

qR

gq

Lb

t1

lL, ν

x4

0,x2+

x3

0

x2

0,x1+

x1

0

840

680

518

380

312

137

116

G

NI, CI

Ninv

28% qq l ν

650

440

100

5% qq ll

60% qq

G

Ninv

700

100

Lep(G) Btag(G)

t

q

qt bqb

W,(Z) l,ν

l,νW

5% qq Z

qq,ll,νν

63% tt34% qq3% bb

~

~~ ~

~~ ~

Complicated exampleFitting examples


Fitting examples

OSO

F

OSSF

Z cand

SSOF

SSSF

Most distributions/counts still well describedDeviations hint at what might be missing in description

One-lepton pT


How to use the simplified models?

● Fit parameters give direct information– Mass scales and mass differences

– Presence of weak bosons or lepton partners

– Over- or underrepresentation of heavy flavor

● Inspiration / starting point for model building– Deviations from data indicates missing features of

the simplified models – for second iteration

● Allows direct comparison with theory using tools available to theorists


How to use the simplified models?

Experimental fit of Simplified model

over SM background

Theoretical comparison Simplified model

vs. three SUSY models

From experimental note Comparison by theorist


Conclusions

● How present/analyze limits and excesses?● Model-independent approaches:

– Simplified models for analysis, characterization and presentation of early excesses

– (Not discussed: Cross-section grids for limits)

● Avoids model-bias – can fit any model● Allows detector-independent comparison with

any theoretical model● Attempt to maximize usefulness of experimental

analyses for the HEP community!


Backup slides


Cross section limits

Example: Non-unified/non-standard SUSY scenarios can have m

g:m

B ~ 1

● Then, gluino decays to 2 soft jets and LSP– No hard jets, no large missing transverse energy

– Need initial state QCD radiation

● A priori unclear whether Tevatron is sensitive– Need combination of MET+1-jet, 2-jet, 3-jet and

multijet searches to cover whole g-B mass plane

● Very difficult to find limits outside collaboration

~ ~

~~



Our suggestion:● Provide differential cross section limits for phase

space bins (in relevant variables) for mutually exclusive searches (1j+MET, 2j+MET, ...)

● Provide detector simulation and event generation chain verified to allow comparison with presented experimental cuts



Our suggestion:● Provide differential cross section limits for phase

space bins (in relevant variables) for mutually exclusive searches (1j+MET, 2j+MET, ...)



● Then, easy for theorists to generate the corresponding model cross sections and compare, point-by-point in parameter space, to get exclusion region.



mSUGRA mass ratiog → qqB

g → qqW+B

~ _~

~ _ ~mSUGRA exclusion

● Then, easy for theorists to generate the corresponding model cross sections and compare, point-by-point in parameter space, to get exclusion region.

model-independent approaches to missing energy at the...

Documents