Search for Dark Matter in Mono-Photon events with ATLASphd.fisica.unimi.it/assets/Seminario-Ratti.pdf · M.G. Ratti - Search for Dark Matter in mono-photon events with ATLAS PhD Workshop

+ Search

for Dark Matter in Mono-Photon

events with ATLAS

Maria Giulia Ratti In collaboration with Silvia Resconi, Leonardo Carminati, Donatella Cavalli

F IRST YEAR PHD WORKSHOP - MILANO

OCTOBER 12 T H , 2015

+ Why Dark Matter at the LHC ?

¤ Compelling evidence of Dark Matter from astrophysical probes

¤ But what is the nature of the Dark Matter? How does it interact with Standard Model particles ?

¤ Complementary strategies for the detection of DM particles:   Direct searches

  Indirect searches

  Production at colliders

2

M.G. Ratti - Search for Dark Matter in mono-photon events with ATLAS PhD Workshop 2015

¤ Grounding assumption:   DM and SM interact other than gravitationally, otherwise none of the

strategies is effective

+ Mono-X Signatures

¤ DM goes out undetected

3


+ Mono-X Signatures

¤ DM goes out undetected => need a visible SM particle to tag the event

¤ Mono-X signatures: ETmiss + X=jet,γ,W, Z, H

4


+ Mono-X Signatures

¤ DM goes out undetected => need a visible SM particle to tag the event

¤ Mono-X signatures: ETmiss + X=jet,γ,W, Z, H

¤ X object irradiated by the initial state or, in the case of electroweak bosons, involved in the interaction

5


q

q̄

γ

γ

χ

χ̄

+ Mono-Photon Search ¤  Cut & Count analysis ¤  Look for a deviation in data from prediction of

the SM => Signal over Background

¤  Essential a very accurate background estimation

¤  Quantify the level of agreement/disagreement by means of a statistical analysis

¤  Set limits on parameter space of various models

6


Which SM processes have the same signature ?

Z(àνν) + γ irreducible background O(70%) W(àlν) + γ O(15%)

W/Z + jets electron or jet taken as a photon O(15%)

γ + jets and other remaining bkgs O(1%)

How are they estimated ?

from DATA/MC RATIOS in appropriate Control Regions

purely DATA-DRIVEN techniques

pure MC simulation

+ Mono-Photon Search 7


Signal Region (SR) •  Trigger and Event cleaning •  Jet cleaning •  Energetic photon with pT > 150 GeV •  ET

miss>150 GeV •  Δϕ( γ, ET

miss) > 0.4 •  leading photon “tight”, isolated •  at most one jet well separated from the

ETmiss

•  Veto on electrons and muons

Control Regions (CRs) •  keep the same cuts as SR

•  revert one or more cuts at a time to define regions enriched in a particular source of background

W/Z + γ Backgrounds

SR Z+γ W+γ

2μ CR Z(μμ)+γ

1μ CR W(μν)+γ

2 ele CR Z(ee)+γ

  Data/MC ratios in the CRs

  Extrapolate to the SR

  Normalize the yields in the SR

+

  Two-dimensional Sideband method

  i

Mono-Photon Search 8


Signal Region (SR) •  Trigger and Event cleaning •  Jet cleaning •  Energetic photon with pT > 150 GeV •  ET

miss>150 GeV •  Δϕ( γ, ET

miss) > 0.4 •  leading photon “tight”, isolated •  at most one jet well separated from the

ETmiss

•  Veto on electrons and muons

Control Regions (CRs) •  keep the same cuts as SR

•  revert one or more cuts at a time to define regions enriched in a particular source of background

Non-Iso

Isolated

Jets faking Photons

1ele CR no photon

EFR

Electrons faking Photons

  EFR = probability of electron to fake a photon with a Tag & Probe method

  Scale a mono-electron CR with this probabulity

+ Data Analysis in ATLAS ¤ A huge and continuously evolving framework is needed for

data analysis:   Big amount of data in various formats spread over the grid   Smaller datasets, called “derivations”, optimized for each analysis, to be

replicated at local sites

  Reconstruction software maintained by the Combined Performance groups   Analysis software developed by each analysis group for their needs

9


+ What is ETmiss ?

¤ ETmiss = Missing Transverse Momentum

  Negative vector sum of the transverse momenta of all detected particles

  Global quantity of the event

  In a parton-parton scatter the initial transverse momentum is ~ 0 => Measured imbalance of the total transverse momentum is the handle for the invisible part of the event

PhD Workshop 2015 M.G. Ratti - Search for Dark Matter in mono-photon events with ATLAS

10

Real ETmiss:

  New particles   Neutrinos

Fake ETmiss:

  Miscalibrations   Mismeasurements   Limited detector acceptance   Detector Noise

ETmiss is the discrimina2ng variable for many searches for new physics

+ ETmiss Reconstruction: Ingredients

  Reconstructed and calibrated “physics objects”:

o  electrons, photons, taus, muons n  Selected as recommended from the various CP groups n  analyses can optimize the selections for their needs

o  Jets: n  Fully calibrated Anti-kt4 with pT > 20 GeV n  Anti-kt4 with pT > 7 GeV for handling the overlap between physics

object

  Signal objects: o  tracks and clusters


11

Muon

Muon

Jet

Jet

Track Cluster

Jet

+ ETmiss Reconstruction: Term by Term

Ex(y)miss = Ex(y)

miss, e + Ex(y)miss, γ + Ex(y)

miss, τ

+ Ex(y)miss, jets + Ex(y)

miss, μ + Ex(y)miss, Soft

Ex(y)miss, k = - Σk px(y)

k


12

Jet

Hard Terms

Soft Terms

  Reconstructed and calibrated “physics objects”: o  electrons, photons, taus, muons

n  Selected as recommended from the various CP groups n  analyses can optimize the selections for their needs

o  Jets: n  Fully calibrated Anti-kt4 with pT > 20 GeV

  Tracks from primary vertex => TST ET

miss   Unmatched clusters + soft jets => CST ET

miss

+


13

[GeV]missTE

0 50 100 150 200 250

Even

ts

1<10

1

10

210

310

410

510

610 missTCST E

missTTST E

missTTrack E

µµ AZ all jets = 13 TeV 25 nss

ATLAS Simulation Internal

[GeV]missTE

0 50 100 150 200 250

Even

ts

1<10

1

10

210

310

410

510 missTCST E

missTTST E

missTTrack E

µµ AZ 0 jet = 13 TeV 25 nss


[GeV]missTSoft Term E

0 20 40 60 80 100 120 140 160 180 200

Even

ts

1<10

1

10

210

310

410

510 missTCST E

missTTST E

missTTrack E

µµ AZ 0 jet = 13 TeV 25 nss


[GeV]missTSoft Term E

0 20 40 60 80 100 120 140 160 180 200

Even

ts

1<10

1

10

210

310

410

510

610 missTCST E

missTTST E

missTTrack E

µµ AZ all jets = 13 TeV 25 nss


  TST ETmiss performing best

  Higher value of the ETmiss and Soft Term at higher jet multiplicity

ETmiss Performance

+ ETmiss Performance: Resolution


14

PVN0 5 10 15 20 25 30

Res

olut

ion

[GeV

]m

iss

y,E

mis

sxE

0

5

10

15

20

25

30

35

40missTCST EmissTTST E

missTTrack E


= 13 TeV 25 nssµµ A Z

all jets

PVN0 5 10 15 20 25 30

Res

olut

ion

[GeV

]m

iss

y,E

mis

sxE

0

5

10

15

20

25

30

35

40missTCST EmissTTST E

missTTrack E


= 13 TeV 25 nssµµ A Z

>20GeVT

0 jets p

  Width of the Ex,ymiss is a sensitive quantity to pile-up effects

  Measured as a function of the number of primary vertices, NPV, in 0-jet (left) and inclusive jet (right) topologies

  In 0-jet events TST and Track ETmiss perform very similar, insensitive to pile-

up   In inclusive jet events, CST and TST ET

miss both depend on pile-up   Track ET

miss stable wrt pile-up

+ ETmiss Performance: Scale in Z è ll

  Well calibrated ETmiss in Zèll events

projected along any axis should be zero

  Projection of the ETmiss onto the Z axis is

sensitive to the imbalance between Hard and Soft part of the event


15

[GeV]ZT

p0 10 20 30 40 50 60

> [G

eV]

ZA u

mis

sTE

<

50<

40<

30<

20<

10<

0

10


= 13 TeV 25 nssµµ AZ

>20GeVT

0 jets p

MET_CST

MET_TST

MET_TRACK

[GeV]ZT

p0 20 40 60 80 100 120 140 160 180 200

> [G

eV]

ZA u

mis

sTE

<

100<

80<

60<

40<

20<

0


= 13 TeV 25 nssµµ AZ

all jets

MET_CST

MET_TST

MET_TRACK

  Bias is bigger in 0-jet events indicating underestimation of the Soft Term   In events with jets, TST ET

miss performs best

+ Conclusions and Plans

¤ Analysis of mono-photon events in good progress:   Most analysis tools in place: derivations, software, selections and

background estimation techniques

  Big effort in understanding the ETmiss, invisible part of the collision

event

  Work in progress in the statistical interpretation of the results

  Almost 2.5 fb-1 of data already collected by ATLAS and ready for analysis

=> will need the entire 2015 data to improve the Run 1 results

16


+

BACK-UP


17

+ Background projections with 5 fb-1


18

[GeV]a

Tp

200 300 400 500 600 700 800 900 1000 1100310×

1<10

1

10

210

Signal Region cuts

aZZ + jetsW + jets + jets and dijetsaaW

a + iiAZ

SimulationATLAS= 13 TeVs, -1 Ldt = 5.0 fb0

[GeV]missTE

200 300 400 500 600 700 800 900 1000 1100310×

1<10

1

10

210

Signal Region cuts

aZZ + jetsW + jets + jets and dijetsaaW

a + iiAZ

SimulationATLAS= 13 TeVs, -1 Ldt = 5.0 fb0

+ Mono-Photon Studies 19


[GeV] aTE

100 120 140 160 180 200

Effi

cien

cy

0

0.2

0.4

0.6

0.8

1

selectionmissTPhoton + E

-1Ldt = 71 pb0 = 13 TeV, s

Data 2015

ATLAS Preliminary

HLT_g120_loose

[GeV]missTE

0 200 400 600 800 1000120014001600180020005<10

4<10

3<10

2<10

1<10

ATLAS Work in Progressa)+iiAZ(

muons invisiblea)+µµAZ(

electrons invisiblea ee)+AZ(

[GeV]missTE

0 100 200 300 400 500 600 700 800 900 10000

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Z+gamma Tight Selection, Tight Ph in MET

Z+gamma Non-Tight Selection, Tight Ph in MET

Z+gamma Tight Selection, Loose Ph in MET

Z+gamma Non-Tight Selection, Loose Ph in MET

>µ<0 5 10 15 20 25 30 35 40

Jet V

eto

Effic

ienc

y

0.50.550.6

0.650.7

0.750.8

0.850.9

0.951

> 30 GeVT

p > 40 GeV

Tp

> 50 GeVT

p

+ TST Soft term systematics

  Systematics on the ETmiss measurement quantify the level of

agreement between data and MC

  Component originated from the measurement of the other physics objects can be propagated through the ET

miss computation

  TST Soft term uncertainties are provided on MC-based studies

  Expected to cover discrepancies between data and MC at 13 TeV

o  Modelling of the generators

o  Full vs Fast simulation

o  Experimental conditions: geometry of the detector, bunch spacing …


20

+ TST Soft term systematics

Soft Term projections onto pThard:

o  Mean of longitudinal component

=> scale uncertainty

o  RMS of transverse and longitudinal components

=> resolution uncertainty


21

+ ETmiss Performance: Resolution


22

²  W è lν inclusive jets (left), ttbar inclusive jets (right)

²  In inclusive jet events, all variants suffer by the increased event activity in higher pile-up regions

²  TST and CST show similar values of the resolution among various topologies, while Track resolution suffers in high-jet multiplicity events

+ TST ETmiss : Resolution and Scale in

2015 data and MC


23

(event) [GeV]T EY

0 100 200 300 400 500 600 700

RM

S [G

eV]

mis

sy

,Em

iss

xE

0

5

10

15

20

25

30 µµ AMC Z

Data 2015

ATLAS Preliminary

-1Ldt ~ 6 pb0 = 13 TeV, s

[GeV]ZT

p0 5 10 15 20 25 30 35 40 45 50

> [G

eV]

ZA u

mis

sTE

<

15<

10<

5<

0

5

10

15ATLAS Preliminary

-1Ldt ~ 6 pb0 = 13TeV, s

µµ AMC Z

Data 2015

²  Agreement between data and MC with very first data, Z è μμ events

+ Jet Selection in ETmiss


24

²  Higher pT threshold for jets going into the jet term can improve the resolution at high NPV

²  but also increases the bias at all pTZ

PVN0 5 10 15 20 25 30

Res

olut

ion

[GeV

]ym

iss

,Exm

iss

E

0

5

10

15

20

25

30

35

40

ATLAS Simulation Preliminary

= 13 TeVs 25nsµµ AZ

all jets

20GeV25GeV30GeV

[GeV]ZT

p0 20 40 60 80 100 120 140 160 180 200

[GeV

]� Z

Aum

iss

T E�

-14-12-10-8-6-4-2024

ATLAS Simulation Preliminary

20 GeV25 GeV30 GeV

= 13 TeVs 25nsµµ AZ

all jets

+ ETmiss Reconstruction: Association Map

⇒  There must be a mechanism to keep track of the overlaps between physics/signal objects:

  Run 2 Association Map: o  Contains the spatial association of each physics object to anti-kt4 jets o  Within each jet, object overlaps are identified o  Unassociated tracks/clusters go into the core soft terms


25

Analysis phys obj

An2-‐kt4 jets

SKL

HWD

8QDVVRFLDWHG�WRSRFOXVWHUV�DQG�WUDFNV

-HW

7DX

(�JDPPD�6OLGLQJ�ZLQGRZ�FOXVWHU

7RSRFOXVWHU,'�WUDFN

Documents

Search for Dark Matter in Mono-Photon events with ATLASphd.fisica.unimi.it/assets/Seminario-Ratti.pdf · M.G. Ratti - Search for Dark Matter in mono-photon events with ATLAS PhD Workshop