Measurement of the Higgs boson coupling with tau leptons and search for an additional neutral MSSM Higgs boson with the ATLAS detector

Measurement of the Higgs boson coupling with tau leptons and search for an additional neutral MSSM

Higgs boson with the ATLAS detector

Damián Álvarez Piqueras

Thesis defense

7th May 2018

directed by Luca Fiorini

1

Outline

1. Summary of Thesis

2. The Standard Model and the Higgs mechanism 3. The LHC and the ATLAS detector

4. The SM H → analysis

5. An extension beyond the SM: MSSM

6. The MSSM A/H → analysis

7. Conclusions

Thesis defense

PhD studies started in October 2012

ATLAS author with qualification task in the

Enhancing of the Pulse Simulator tool

Technical tasks for the maintenance and functioning of the TileCal detector, such as shifter in control room and DQ

Contributions to two analyses related to the Higgs to channel that led to two published papers

Small contribution in outreach such as talks, events, expositions and student visits

3

The Standard Model and the Higgs mechanism

The Standard Model

and the Higgs boson

The Standard Model (SM) is the

theory that describes the behaviour of subatomical particles and their

interactions:

- Fermions

- Quarks (uct, dsb)

- Leptons: electron, neutrino-type

- Bosons - interactions - Strong int (QCD): g

- Electroweak int: W, Z, - Scalar: Higgs boson

- Problem: W and Z are massive

- Mass 〜 resistance to movement

- Solution: Higgs mechanism

5

The Higgs mechanism

- A new undiscovered field is proposed: the Higgs field

- The field has a feature: non-zero expectation value for vacuum

- Even in vacuum, there is remanent energy, in the form of particles: Higgs bosons

- When particles propagate, they interact to the Higgs boson

- The higher the coupling, the less the particle can propagate:

- 〜 appearance of resistance to movement : “mass”

- The “mass” of the fundamental particles is proportional to its coupling to the Higgs boson

- , g do not interact → massless - W, Z do interact → massive

- Quarks and leptons also interact with Higgs boson → Yukawa coupling

- coupling proportional to their mass

The Higgs boson

- The Higgs boson was discovered in 2012 in the bosonic channels at

m

〜 125 GeV

- H →

- H → ZZ → 4l

-

H → WW

- Next step → direct fermionic coupling!

H →

- Relevant branching ratio ( 〜 7%)

- Easy to detect against background

7

The LHC and the ATLAS Experiment

The Large Hadron Collider at CERN

- CERN is the European Organization for Nuclear Research - The LHC is a 27-km-long proton collider

- A set of machines accelerate protons in different stages to the LHC - Two parallel beams of protons at specific energy cross

- Four experiments are placed in the colliding points

9

ATLAS detector

- Multi-purpose detector at P1 of LHC - Designed to search for the Higgs

boson and BSM at TeV scale

- Composed of different subdetectors:

- Inner Detector

- Run 2: ID + IBL

- Liquid Argon Calorimeter (ECAL) - Tile Calorimeter (HCAL)

- Muon Spectrometer - Central solenoid

- Barrel + EndCap Toroids - Trigger + DAQ system

- Forward detectors ¹⁰

- The SM analysis used the data collected during the full Run 1 of LHC

- An integrated luminosity of L = 4.5 fb⁻¹ at e.c.m 7 TeV in 2011 - An integrated luminosity of L = 20.3 fb⁻¹ at e.c.m. 8 TeV in 2012

- The MSSM analysis used the data collected during the first year (2015) of the Run 2:

- An integrated luminosity of L = 3.2 fb⁻¹ at e.c.m. of 13 TeV and bunch-spacing of 25ns

-

Datasets

Run 1

Run 2 11

The SM H → analysis

First analysis: SM H →

- Objective: Measurement the coupling of the Higgs boson to leptons and comparison it with the SM expectation

- Methodology: Estimation the number of di-tau events expected by the SM (without Higgs) and comparison it with observation

- How can the Higgs boson be produced at LHC?

- How does the Higgs boson decay?

- What other processes can be reconstructed as Higgs boson?

13

- The Higgs boson is produced by three main processes - gluon-fusion (ggH): most abundant process ( 〜 85%)

- Vector Boson Fusion (VBF):

- Vector-associated production (VH): small contribution ( 〜 5%)

Higgs production

ggH VBF VH

VH

ggH VBF WH ZH

σ x BR [pb] 1.22 0.100 0.0445 0.0262

Order NNLO+NNLL (N)NLO NNLO NNLO

14

- The Higgs boson decays in two opposite-charged tau leptons

- The tau leptons also decay into a tau-neutrino ( ) mediating a W boson - According to the decay of the W boson, there are three final states:

- Fully hadronic (

) – 〜 42%

- Semi-leptonic (

) – 〜 45.6%

- Fully-leptonic (

) – 〜 12.4%

- Two light leptons (e, ) and four undetectable neutrinos

Higgs decay

H → ^–→ _l

l

Higgs decay: H → lep lep

Two light leptons (e, ) and four undetectable neutrinos

- Visible mass (m

): Invariant mass of the di-lepton system

- Collinear mass (m ): computed assuming the collinear approximation

- The taus are boosted, so their decay products move in same direction

- The system of unknown variables can be resolved if taus are not back-to-back - Since tau are considered massless, approximation is reasonable valid

- Missing Mass Calculator (MMC): algorithm that computes the mass by estimating the exact direction of the neutrinos

- Not all topologies are equally probable

- Using extra info as PDFs e.g. ∆R( ) from Z → process - MMC depends greatly on the E_T^miss accuracy and resolution

Reconstruction of di-tau mass

17

- Main background:

- neutral boson, similar mass - Irreducible background

- Estimation with data-driven technique:

EMBEDDING - Sample of data Z →

- are substituted by MC + decay -

, jets, UE,.. are modeled by data - Tau kinematics modeled by MC

- Embedding is normalized to MC in an early stage

Background: Z →

- Contribution of two light leptons without relevant

- Only relevant in Same Flavour events (SF)

- Estimation with MC

- A Control Region (CR) for this bkg is defined under the Z peak

- A

correction is derived from CR and applied to Signal Region (SR)

Background: Z → ll

19

- tt and (single-top) can be source of two light leptons - Events with high

and several jets, especially b-jets - Estimation with MC

- Background is reduced by vetoing events with a b-tagged jet - A CR for this bkg is defined asking for a b-jet

Background: Top-quark processes

- di-boson process (WW, ZZ, WZ) can be source of two light leptons - They have small cross-section

- Estimation with MC and controlled in a Validation Region (VR)

Background: di-boson and H → WW

- Special case of di-boson: H → WW - They are considered background

with SM cross-section - A cut is implemented to

orthogonalize between and WW analyses

- by requiring m > m

– 25 GeV

- Some non-lepton objects can be mis-reconstructed as leptons → fake - Mainly events where one lepton is faked (W+jets, semi-leptonic tt) - Estimation is done using a data-driven technique:

Background: Fake background

Template Fit

- Template obtained in multi-jet enriched CR

- CR is defined by inverting isolation criteria - True-lepton bkg in CR are estimated with

MC and subtracted

- Normalization is obtained by fitting the template of p_T (L2) distribution from CR - Same process is done in same-sign (SS)

events to act as validation (VR)

- A systematic uncertainty is derived from

comparison SS-OS ²¹

➢ Aim: to reduce background and enhance signal over bkg ratio - Pre-selection: event cleaning of non-collision or faulty event - Selection in leplep: trigger of two-lepton events

1. two isolated leptons with opposite charge, veto of hadronic tau

2. m_ll in range [30 - 75] GeV for SF and [30 - 100] GeV for DF (Zll bkg) 3. high energetic leptons: p_T(L1) + p_T(L2) > 35 GeV (Fakes bkg)

4. at least one jet with p_T > 40 GeV (enhance boosted topology)

5. presence of : (E_T^miss, HP_TO) > 40 GeV for SF, E_T^miss > 40 GeV for DF (Zll + Fakes) 6. Collinear approximation: x₁ and x₂ are in range [0.1 - 1]

7. ∆ (LL) < 2.5 (Zll and tt bkg) ← MAIN SELECTION 8. Categorization

- VBF - Boosted 9. Final selection

- veto of b-jets (tt bkg)

- Orthogonalization against H → WW: m > m_Z – 25 GeV - Physical solution to MMC: MMC > 0

Event selection: Signal Region (SR)

➢ Aim: split dataset in events with different topologies → better sensitivity

●

Categorization

VBF

- Aimed for VBF events

- Clear signature: two jets back-to-back in the beam direction

- Most sensitive category

- Requirements:

● At least two jets

● with p_T(j1) > 40 GeV, p_T(j2) > 30 GeV

● Angular separation ∆ (jj) > 2.2

Boosted

- Events with boosted topology

● ggH process with one jet from initial-state radiation

● Higgs boson is boosted by recoil

- Requirements:

● Non-VBF event

● p_T( ) > 100 GeV

---

Rest of events are discarded

23

- Standard way: sequential cuts

- Problems: cut is absolute, lose signal in each cut

- New approach: a Multivariate Algorithm -

-

-

-

-

Signal Enhancing: BDT

Features:

- Choose an appropriate set of variables (big separation power)

- Optimize parameters of algorithm - Train algorithm to recognize patterns - Check consistency of results

Signal Background

-

:

- VBF: MMC, ∆R (LL), |∆ (jj)|, min(∆ (LL, jets)), m_JJ, C (L1) x C (L2), ∆ (j3,jj)

-

-

- Optimization of parameters

- 2D scans of configuration parameters (error fraction), statistical sensitivity

- Check consistency of results: ROC curve, KS tests,...

- Cut-based analysis to cross-check

- Training

- Cross-evaluation to avoid biases

- Signal: VBF for VBF cat and ggH for Boosted cat - Fakes: no MC (negative weights)

- H → WW process not taken into account

BDT details

25

- Value in range [-1, +1] representing compatibility of event with Bkg (-1) or Signal (+1) - BKG: If there is agreement in low-score region, bkg is well modelled

- SIG: signal is concentrated in last bins, expected as an excess of events over bkg - Here, signal is normalized, excess will be much smaller

BDT Score

- BDT Score distribution in SR for VBF and Boosted cats

● BDT Score: value in range [-1, +1] representing compatibility of event with Bkg (-1) or Signal (+1)

● BKG: If there is agreement in low-score region, bkg is well modelled

● SIG: signal is concentrated in last bins, expected as an excess of events over bkg

Results lep lep channel: Signal Region

VBF: BDT Score SR Boosted: BDT Score SR ²⁷

- MMC variable in SR for VBF and Boosted cats

● Excess of events expected ∼ 125 GeV

Results lep lep channel: Signal Region

Boosted: MMC SR ²⁸

VBF: MMC SR

An excess of data over the background is observed with a observed statistical significance of:

4.5

Results SM H →

MMC SR Weighted S/B events

SR and CR distributions are fitted to bkg and signal expectation to extract signal strength :

value which is compatible with the SM expectation ²⁹

Results SM H → : Compatibility

CUT-BASED MVA 2D-contour

Signal strength obtained for the production modes:

ggH and vector-mediated (VBF+VH).

Best fit (red cross) is compatible at 1 with SM expectation (blue cross), and significantly distant from the null-hypothesis, (black cross)

A cut-based approach (optimize to mirror MVA analysis) is used as a cross-check of the main MVA result

Result is compatible with the MVA analysis

31

Beyond the Standard Model: MSSM

- The SM is not a complete theory

● does not provide explanation for several natural phenomena

● some of its mechanisms do not seem

“natural” (hierarchy, fine-tuning) - Several extensions have been proposed - One of them is supersymmetry (SUSY)

● Additional symmetry fermion/boson

● New set of particles to be discovered

● Its minimum implementation is MSSM The Higgs mechanism in the MSSM

- The Higgs mechanism has to be modified - H [ h, H , A, H ⁺, H ^– ]

- H and A could be found at LHC!

● And is a favored search channel!

- discovered in 2012 - neutral scalar

- neutral pseudo scalar

- Charged Higgs

Phenomenology of the MSSM Higgs signal

The MSSM has some modifications respect to SM

- Coupling of the Higgs bosons to fermions is weighted by one factor: tan - At first order, MSSM depend only on tan and m_A parameters

- To reduce other dependences, higher-order parameters are set in such way that enhance certain phenomenologies, defining benchmark scenarios

● m_h^mod : modification (+/-) of s-top mixing param. to make m_h the 125 GeV boson

● hMSSM: the m_h is set to the SM boson value

● light-stau, light-stop, tauphobic,....

Why the search channel is favored?

- Coupling to down fermions ( , b) is enhanced for large values of tan

- Thus, A/H → coupling is enhanced!

- Also, bbH production mode becomes relevant

- And ggH could include b-quark loop

bbH ggH

33

The MSSM A/H → analysis

Second analysis: MSSM A/H →

- Objective: Search for additional MSSM Higgs bosons

- Methodology: Look for excess of di-tau events respect to SM expectation in the high mass range [200 - 1200 GeV]

- How these Higgs bosons behave compared to SM?

- What other processes can be reconstructed as Higgs boson?

- True backgrounds: Z→ , top quark, di-boson - Fake backgrounds: W→ + jets, multi-jet

- For this analysis we consider two final states:

- Fully-leptonic ( _{lep lep}) – 〜12.4%

- Semi-leptonic ( _{lep had}) – 〜45.6%

- Fully hadronic ( _{had had}) – 〜42%

- Two hadronic taus and two undetectable neutrinos

–→

(qq)

35

- Different algorithms to compute the di-tau mass

- Total transverse mass:

where:

- Missing Mass Calculator (MMC)

:

- algorithm that computes the mass by estimating the exact direction of the neutrinos with PDFs from topologies of decays

- MMC was optimized for high-mass analysis

- Other algorithms: MOSAIC and visible mass

- A study was performed to choose the best discriminant

- Figure of merit was the integrated, binned, Asimov significance Z - The total transverse mass (m_T^tot) was chosen as the

final discriminant of the analysis

Reconstruction of di-tau mass

36

- Main background in the _{had had} channel: multi-jet

- One or two jets are mis-identified as a

- Very large cross-section and low prob. of passing selection: no feasible MC - A data-driven technique is implemented: Fake Factors

Fake Factor (FF)

- Factors derived in a di-jet enriched CR to estimate background in SR

● CR: single-jet trigger, two _had with p_T > 55 GeV, and p_T balance (30%), p_T1 fails medium ID and p_T1 > 100 GeV

● Tag and probe analysis: factors = ratio of 2 that pass/fail the -ID in the CR

● Factors parametrized as function of

p_T, prongs, OS/SS and category → OS/SS combined!

● Estimation in SR: FF applied to estimation in region of fail 2-ID

Background modeling: Fake background

37

backup

Background modeling: Other backgrounds

Fake Rate technique (FR)

- Factors derived in a CR to correct background in SR

● Single-muon trigger, p_T > 55 GeV, one _had with p_T > 50 GeV and ∆ ( , _had) > 2.4

- Applied on the MC for each reconstructed not matched to trigger

- Parametrized in terms of p_T and number of prongs, for OS and SS, and lead/sublead - Derived independently for the two categories

● Derived in b-tag CR (enriched with tt): applied to Top processes (tt and single-top)

● Derived in b-veto CR (enriched with W→ ): applied to remaining processes

- Other processes:

- Their contribution is estimated using MC samples

- However, these processes can also have a jet mis-identified as _had - Probability of this is modelled with data-driven technique: Fake Rate

- An additional weight (from a W→ CR) is applied to W→ events as a function of m_T^tot

- Aim: to reduce background and enhance signal over bkg ratio - Pre-selection : cleaning of non-collision or faulty events

- Selection in hadhad: single- trigger with p

> 80 GeV

1. The event contains at least two _had candidates

2. Leading _had passes ‘medium’ -ID and has p_T > 110 GeV 3. Sub-leading _had passes ‘loose’ -ID and has p_T > 55 GeV 4. Leading _had is matched to the object that fired the trigger

5. Veto of any electron or muon objects ( defined as passing ‘loose’ criteria) 6. Back to back topology: ∆ ( ₁, ₂) > 2.7

7. The two _had candidates have opposite electric charge ← Signal Region 8. Categorization conditions:

- b-tag category: focused in bbH, requires events with at least one b-tagged jet - b-veto category: focused in ggH, vetoes events with at least one b-tagged jet

In addition, a VR is defined for each category by inverting the charge requirement

- Same Sign VR

Event Selection

39

Distributions of m_T^tot for each category in the Same Sign VR - Reasonable agreement within uncertainty

Results of the had had : Validation Region

SS b-tag SS b-veto

Distributions of m_T^tot for each category in the Signal Region

- No excess of events over the background expectation is observed

Results of the had had : Signal Region

SR b-tag SR b-veto

41

- No excess of data over the background model is observed - Limits (95% CL) are extracted, using two approaches:

- Model-independent: Limits on the x BR of the production modes

- Model-dependent: Limits on the tan - m_A parameter space of different MSSM scenarios

Results of the A/H → analysis: Limits

LH in backup

ggH bbH

- No excess of data over the background model is observed - Limits (95% CL) are extracted, using two approaches:

- Model-independent: Limits on the x BR of the production modes

- Model-dependent: Limits on the tan - m_A parameter space of different MSSM scenarios

Results of the A/H → analysis: Limits

LH in backup

m_h^mod+ scenario hMSSM scenario 43

Conclusions

Summary

J. High Energy Phys. 04 (2015) 117 Eur. Phys. J. C 76 (2016) 28 45

Current status of the searches

SM H →

ATLAS and CMS combined the Run 1 data

with an obs. (exp.) significance of

5.5

MSSM A/H →

Paper published with first Run 2 data (3.2 fb^-1) The result improves previous searches for the mass range m_A > 500 GeV

Later analysis were done for ICHEP’16 (13.3 fb^-1) and 2015+2016 luminosity (36.1 fb^-1)

Conclusions

- This thesis was performed during the Runs 1 and 2 of the LHC, and the Long Shutdown, both in physics analyses and technical tasks of the Tile Calorimeter of the ATLAS detector

- The work of this thesis has contributed to obtain strong evidence of the direct coupling of the Higgs boson to fermions, a result which is fully compatible with the Standard Model

- In addition, it has contributed to expand the non-allowed space of

parameters of one extension beyond the Standard Model, the MSSM, and to prepare the field to the upcoming data of Run 2

47

Thank you for your attention!

Backup

49

Enhancement of the Pulse Simulator

Simulation of high pile-up conditions

Old version: one out-of-time pulse

51

New version: seven out-of-time pulses (bunch-space configurable)

- Previous version, only one out-of-time pulse was allowed

- Improvement: added one additional pulse per bunch-crossing

Extended pulse shape

Pulse shape Blue → old version Red→ full available

52

New version: pulses with full shape

- Previous version: Pulse shape only implemented in range [-75ns, +75ns]

- Improvement: Implemented all available range [-75ns, +130ns]

- This allows to study effect of more ‘distant’ pulses and study set-back phase

(negative values for amplitude for t > 100ns)

Old version:

HG → mis reconstructed saturated pulses LG → excess of events

Implementing a mono-gain system

53

- Previous version: PS was implemented in bi-gain system → not correct behaviour - Improvement: Implemented gain-switch and mono-gain system

- Same behavior as real detector

- Better handling of saturating event (no duplicated) and mis-reconstructed

New version:

HG → No mis-reco saturated pulses LG → only saturating pulses

SM H → analysis

Object reconstruction

- Particles produced during collisions leave traces in different detectors

- Several algorithms are used to combine information to reconstruct the original particle that produced the detector response

Reconstruction of particles for the analyses

- Tracking and vertexing: points in silicon detectors are used to generate tracks.

Iterative combination of tracks is used to reconstruct the vertices

- Electrons (e): energy deposits in EM calorimeter and matching track. Cut-based (Run 1) and likelihood discriminant (Run 2) are used to set efficiency working points.

- Muons ( ): track segments from ID and Muon Spectrometer - Jets: calibrated clusters of energy deposits in calorimeter

a special algorithm tags the jets coming from a b-quark

- Tau ( ): only visible part of hadronic-decaying is reconstructed as stand-alone object Seed from jets, calibrated, with charge (±1) and prong (1,3) conditions

The -ID is done with a BDT with 3 working points (tight, medium, loose) - E_T^miss: neutrinos ( ) are reconstructed as missing energy in the transverse plane

55

- Used the data collected during the full Run 1 of LHC

- An integrated luminosity of L = 4.5 fb

at e.c.m 7 TeV in 2011 - An integrated luminosity of L = 20.3 fb

at e.c.m. 8 TeV in 2012.

Datasets

Accumulated luminosity Number of interactions per

bunch-crossing ⁵⁶

- Used the data collected during the first year (2015) of the Run 2 of LHC - An integrated luminosity of L = 3.2 fb

Datasets

Accumulated luminosity Number of interactions per

bunch-crossing ⁵⁷

- The Higgs boson is produced by three main processes - gluon-fusion (ggH): most abundant process ( 〜 85%)

- Vector Boson Fusion (VBF):

- Vector-associated production (VH): small contribution ( 〜 5%)

Higgs production

ggH VBF VH

VH

ggH VBF WH ZH

σ x BR [pb] 1.22 0.100 0.0445 0.0262

Order NNLO+NNLL (N)NLO NNLO NNLO

58

- Visible mass (m

): Invariant mass of the di-lepton system

- Collinear mass (m ): computed assuming the collinear approximation

- The taus are boosted, so their decay products move in same direction

- The system of unknown variables can be resolved if taus are not back-to-back - Since tau are considered massless, approximation is reasonable valid

- Missing Mass Calculator (MMC): algorithm that computes the mass by estimating the exact direction of the neutrinos

- Not all topologies are equally probable

- Using extra info as PDFs e.g. ∆R( ) from Z → process - MMC depends greatly on the E_T^miss accuracy and resolution

Reconstruction of di-tau mass

backup

59

- Contribution of two light leptons without relevant

- Only relevant in Same Flavour events (SF)

- Estimation with MC

- A Control Region (CR) for this bkg is defined under the Z peak

- A

correction is derived from CR and applied to Signal Region (SR)

Background: Z → ll

- tt and (single-top) can be source of two light leptons - Events with high

and several jets, especially b-jets - Estimation with MC

- Background is reduced by vetoing events with a b-tagged jet - A CR for this bkg is defined asking for a b-jet

Background: Top-quark processes

61

➢ Aim: prove the background are well understood and modeled

● Control Regions: also used in the analysis (normalization for the fit)

- Zll CR: same selection as SR but inverting m_ll condition and only in SF channel - Top CR: same selection as SR but inverting b-veto condition

- Ztt VR: Ztt bkg cannot be separated from signal

● A low-contaminated region can be defined using HP_TO, asking m^HPTO( ) < 100 GeV - Di-boson VR: a OS events, jet-free region

Control and Validation regions (CR / VR)

Ztt VR

At main selection

di-boson VR

Background: Control Regions

63

VBF Top CR

Boosted Top CR VBF

Zll CR Boosted

Zll CR

-

Agreement between background model and data!

● Distribution of p

(L2) at different stages!

Control plots of the background modeling

First stage

(cuts 1-6)

Before categorization

(cuts 1-7) ⁶⁴

-

Agreement between background model and data!

● Distribution of p_T (L2) in Signal Regions of VBF and Boosted categories

Control plots of the background modeling

VBF SR Boosted SR

65

-

Agreement between background model and data!

● Distribution of p

(L2) at different stages!

● HERE: Zll CR of VBF and Boosted categories

Control plots of the background modeling

VBF Zll CR Boosted Zll CR

-

Agreement between background model and data!

● Distribution of p

(L2) at different stages!

● HERE: Top CR of VBF and Boosted categories

Control plots of the background modeling

VBF Zll CR Boosted Zll CR

67

MVA analysis

MVA Overtraining

69

Variables definition: Centralities

Variables definition: Sphericity

71

Variables: VBF

Variables: VBF

73

Variables: Boosted

Variables: Boosted

75

Variables: Boosted

MVA: Cross-checks

77

MVA: Cross-checks

MVA: Cross-checks

79

MVA: Cross-checks

MVA: Cross-checks

81

MVA: Cross-checks

MVA: Cross-checks

83

MVA: Cross-checks

Additional Results

85

- BDT Score for SR, Zll CR and Top CR for VBF and Boosted cats

● BDT Score: value in range [-1, +1] representing compatibility of event with Bkg (-1) or Signal (+1)

● BKG: If there is agreement in low-score region, bkg is well modelled

● SIG: signal is concentrated in last bins, expected as an excess of events over bkg

Results of the lep lep channel: Control Regions

VBF: Zll CR

Plots in backup

VBF: Top CR ⁸⁶

- BDT Score for SR, Zll CR and Top CR for VBF and Boosted cats

● BDT Score: value in range [-1, +1] representing compatibility of event with Bkg (-1) or Signal (+1)

● BKG: If there is agreement in low-score region, bkg is well modelled

● SIG: signal is concentrated in last bins, expected as an excess of events over bkg

Results lep lep channel: Control Regions

Boosted: Zll CR

Plots in backup

Boosted: Top CR ⁸⁷

Results channel: BDT Results: LL

Results channel: BDT Results: LH

Results channel: BDT Results: HH

-

Results lep lep channel: Yields

91

-

Results lep had channel: Yields

-

Results had had channel: Yields

93

Results channel: Systematics

Results channel: Results

Results channel: Cut-based

Results channel: Cut-based

97

For 8 TeV only

MSSM A/H → analysis

MMC Optimization

99

MMC Optimization

Example of deviation distribution Events with jets

mH=600, sqrt(sumET) in [32,34]

100

- MET Resolution → Deviation = MET (Reco) - MET (Truth)

- Distributions of deviation for slices of sqrt(sumET), for events with and without jets - Distributions are fitted with gaussian → sigma extracted

- Done for mass points from mH=200 GeV to mH=1400 GeV

Example of deviation distribution Events without jets

mH=600, sqrt(sumET) in [32,34]

MMC Optimization

101

- Sigmas of different slices are plotted together

- Distribution is fitted with linear function, slope is extracted - Fit done for each mass point

MMC Optimization

- Slopes for different masses are reasonable compatible (check table!) - So, repeat study combining mass points

MMC Optimization

103

- Slopes for different masses are reasonable compatible (table!) - So, repeat study combining mass points

MMC Optimization

- Combining mass points: Linear fit and function

MMC Optimization

105

- Comparison of old and new reconstructions

MMC Optimization

Mass discriminant

107

- Comparison of MMC, visible, MOSAIC, mTtot with Asimov Significance

Mass discriminant

Mass discriminant

109

Additional results

Distributions of m_T^tot for each category in the SR - No excess observed

Results of the lep had : Signal Region

SR b-tag SR b-veto

111

-

Results lep had channel: Yields

Distributions of m_T^tot for inclusive category in the SR No significant excess is observed

Results of the Z’ : Signal Region

Z’ lephad Z’ hadhad

113

Results of the Z’ : Signal Region

Results of the : Systematics

115

Results of the : Extra scenarios

mh mod- mh max

Results of the : Extra scenarios

light stau light stop 117

Results of the : Extra scenarios

tauphobic

Summary

J. High Energy Phys. 04 (2015) 117 Eur. Phys. J. C 76 (2016) 28 119