Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
BSM searches at ATLAS
Run-1 Anomalies, Bumps and Deviations
Alex Martyniuk, UCL
November 5, 2015
1 / 35 Alex Martyniuk BSM searches at ATLAS
Introduction – Standard Model @ ATLAS
pp
total
80 µb−1
JetsR=0.4
|y |<3.0
0.1< pT < 2 TeV
DijetsR=0.4
|y |<3.0y ∗<3.0
0.3<mjj < 5 TeV
W
fiducial
35 pb−1nj ≥ 0
nj ≥ 1
nj ≥ 2
nj ≥ 3
nj ≥ 4
nj ≥ 5
nj ≥ 6
nj ≥ 7
Z
fiducial
35 pb−1nj ≥ 0
nj ≥ 1
nj ≥ 2
nj ≥ 3
nj ≥ 4
nj ≥ 5
nj ≥ 6
nj ≥ 7
t̄t
fiducial
e, µ+X
nj ≥ 4
nj ≥ 5
nj ≥ 6
nj ≥ 7
nj ≥ 8
tt−chan
total
WW
total
γγ
fiducial
Wt
total
2.0 fb−1
H
fiducial
H→γγ
VBFH→WW
ggFH→WW
H→ZZ→4ℓ
total(γγ,ZZ )
H→ττ
WZ
total
13.0 fb−1
ZZ
total
Wγ
fiducial
WW+WZ
semilept.fiducial
Zγ
fiducial
t̄tW
total
t̄tZ
total
95% CL
upper
limit
t̄tγ
fiducial
ZjjEWK
fiducial
Wγγ
fiducialnjet=0
W±W±jjEWK
fiducial
ts−chan
total
95% CL
upper
limit
0.7 fb−1
σ[p
b]
10−3
10−2
10−1
1
101
102
103
104
105
106
1011
LHC pp√s = 7 TeV
Theory
Observed 4.5 − 4.9 fb−1
LHC pp√s = 8 TeV
Theory
Observed 20.3 fb−1
Standard Model Production Cross Section Measurements Status: March 2015
ATLAS Preliminary
Run 1√s = 7, 8 TeV
2 / 35 Alex Martyniuk BSM searches at ATLAS
Introduction – The Problems
Things are not as rosy as they seem for the Standard Model (SM)
Problems lurk within.....It contains no dark matter candidateIt requires immense fine tuning for the 125 GeV Higgs to make senseIt does not contain gravity
Beyond the SM (BSM) theories try to solve these issues
The ATLAS SUSY, Exotics and Higgs groups search for a wide range of BSMphysics utilising the full ATLAS detector
Often at the limits or beyond the specification the detector
We built a general purpose detector for a reason!
3 / 35 Alex Martyniuk BSM searches at ATLAS
Introduction – Obligatory ATLAS Slide
4 / 35 Alex Martyniuk BSM searches at ATLAS
Outline – What I will show
I will cover a whirlwind tour of three searches from the ATLAS Exotics/SUSYgroups that had some ‘oddities’ in Run-1
SUSY: Pair produced cascade decays search using same-flavouropposite-sign dilepton pair, jets, and large missing transverse momentum
[SUSY-2014-10]
Exotics: Search for anomalous production of b-jets and a pair of leptons of thesame charge
[EXOT-2013-16]
Exotics: Search for diboson resonances decaying to boson-tagged jets[EXOT-2013-08]*Fair warning, I worked directly on this analysis and might well be biased to spend more time here...
5 / 35 Alex Martyniuk BSM searches at ATLAS
SUSY Searches – needle in the haystack?
6 / 35 Alex Martyniuk BSM searches at ATLAS
SUSY Searches – ExclusionsATLAS SUSY summary papers out now,[SUSY-2014-05], [SUSY-2014-06], [SUSY-2014-07], [SUSY-2014-08]
[GeV]1t
~m
200 300 400 500 600 700 800
[G
eV
]10
χ∼m
0
50
100
150
200
250
300
350
400
450
1
0χ∼ t →
1t~
1
0χ∼ t →
1t~
1
0χ∼/b f f’
1
0χ∼ W b →
1t~
1
0χ∼ W b →
1t~
1
0χ∼ c →
1t~
1
0χ∼ b f f’ →
1t~
1
0χ∼
,t) <
m1t~
m(
∆
W
+ m
b
) <
m1
0χ∼,
1t~ m
(∆
) <
01
0χ∼,
1t~ m
(∆
1
0χ∼ t →1t
~ /
1
0χ∼ W b →1t
~ /
1
0χ∼ c →1t
~ /
1
0χ∼ b f f’ →1t
~ production, 1t
~1t
~
ATLAS
1
0χ∼W b
1
0χ∼c
1
0χ∼b f f’
Observed limits Expected limits All limits at 95% CL
1=8 TeV, 20 fbs
t0L/t1L combined
t2L, SC
WW
t1L, t2L
tc
tc, t1L [GeV]
1t~m
170 180 190 200 210
[G
eV
]10
χ∼m
0
10
20
30
40
) [GeV]q~
m(
300 400 500 600 700 800 900 1000
) [G
eV
]10
χ∼m
(
100
200
300
400
500
600
700
800
900
1000
, x=1/20
1χ∼0
1χ∼ qqWW→ q
~q
~
ATLAS
1=8 TeV, 20 fbs
0L + 1L combination
)1
0
χ∼
) < m
(
q~m(
)theory
SUSYσ1 ±Observed limit (
)expσ1 ±Expected limit (
1lepton alone
0lepton alone
All limits at 95% CL
[GeV]±
1χ∼m
100 200 300 400 500 600 700 800
[G
eV
]0 1
χ∼m
0
100
200
300
400
500
600
0
1χ∼
= m
±
1χ∼m
1=8 TeV, 20.3 fbs
All limits at 95% CL
lCombined SS MVA, 3
0
1χ∼
) ν ν l l (0
1χ∼ ν l →) ν ν
∼l (Ll
~ ν
∼), l ν ν
∼l(
Ll~ ν
Ll~ →
0
2χ∼
±
1χ∼
Simplified Model:
)0
1χ∼ m
±
1χ∼ + 0.95*(m0
1χ∼ = m
Ll~
m
0
2χ∼ = m±
1χ∼m
ATLAS
)theory
SUSYσ1 ±Observed limit (
)expσ1 ±Expected limit (
lObserved limit 3
Observed limit SS MVA
) [GeV]g~
m(
400 600 800 1000 1200 1400
) [G
eV
]10
χ∼m
(
100
200
300
400
500
600
700
800
900
1000
, x=1/20
1χ∼0
1χ∼ qqqqWW→ g
~g
~
ATLAS
1=8 TeV, 20 fbs
0L + 1L combination)1
0
χ∼
) < m
(
g~m
(
)theory
SUSYσ1 ±Observed limit (
)expσ1 ±Expected limit (
1lepton alone
0lepton alone
All limits at 95% CL
Fra
ctio
n of
Mod
els
Exc
lude
d
0
0.2
0.4
0.6
0.8
1
) [GeV]g~m(
0 500 1000 1500 2000
) [G
eV]
0 1χ∼m
(
0
500
1000
LSP0
1χ∼pMSSM:
1−=8 TeV, 20.3 fbs
[1405.7875]0
1χ∼ qq→ g~
ATLAS
Fra
ctio
n of
Mod
els
Exc
lude
d
0
0.2
0.4
0.6
0.8
1
) [GeV]g~m(
0 500 1000 1500 2000)
[GeV
]q~
m(
0
1000
2000
3000
LSP0
1χ∼pMSSM:
1−=8 TeV, 20.3 fbs
)=0 GeV [1405.7875]0
1χ∼m(
ATLAS
7 / 35 Alex Martyniuk BSM searches at ATLAS
SUSY search using ll + jets + EmissT
[SUSY-2014-10]
‘Standard’ ATLAS SUSY search methodologyControl regions (CRs): Used to constrain SM backgroundsValidation regions (VRs): CR background estimates extrapolated into VRs ascross-checkSignal regions (SRs): After VR check, extrapolate background into SR and look forexcess of events over expected backgroundSearch performed over a grid of signal masses
8 / 35 Alex Martyniuk BSM searches at ATLAS
On shell Z channel regions
Backgrounds studiedFlavour symmetric t t̄ , WW , Wt : Data-driven.CReµ data (corrected for other backgrounds).Lepton selection eff. relations transfer to SR.Cross-checked with Z -mass sidebands. MCnorm. in CRT, extrapolation checked in VRT/Z.Z/γ+ jets: Data-driven. Seed CR jets smearedto mimic Emiss
T distribution. Tested in VRZ region.Fake leptons: Data-driven. Derived from matrixmethod.Other backgrounds. WZ , ZZ , Rare top. Modelledby MC.
9 / 35 Alex Martyniuk BSM searches at ATLAS
Flavour symmetric modellingBackground normalisation modelling in goodagreement with the data in the flavour symmetriccross-check CR/VRs (Z/γ+ jets in backup)
High-HT : HT > 600 GeV
High-EmissT : Emiss
T > 225 GeV, 400 > HT > 600 GeV
VRT
(HT > 500 GeV)
���
��� ���
��� ���
CRT
(HT > 600 GeV)
SR-Z
(HT > 600 GeV)
VRT
(HT > 500 GeV)
VRTZ
(HT > 500 GeV)
CRT
(HT > 600 GeV)
� ��� ��� ����� ���
�� ��
Eve
nts
1
10
210
310
Data
Flavour Symmetric
* + jetsγZ/
Other Backgrounds
Total SM
-1=8 TeV, 20.3 fbsATLAS
Z-mass side band Z-mass window
CRT VRT VRT VRT VRTZ VRTZ VRTZ
tot
σ)
/ ex
p -
Nob
s(N -2
0
2
4 µµee+ee
µµ
)T
(high H )miss
T(high E )
T(high H )
miss
T(high E
10 / 35 Alex Martyniuk BSM searches at ATLAS
SR Results
Excesses seen in both channels (ee: 3.0σ, µµ: 1.7σ)
Combined significance 3.0σ
CMS observes no excess
[GeV]llm82 84 86 88 90 92 94 96 98 100
Eve
nts
/ 2.5
GeV
2
4
6
8
10
12
14ATLAS
-1 = 8 TeV, 20.3 fbs
SR-Z ee
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
[GeV]missTE
200 250 300 350 400 450 500
Eve
nts
/ 25
GeV
2
4
6
8
10
12ATLAS
-1 = 8 TeV, 20.3 fbs
SR-Z ee
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
[GeV]llm82 84 86 88 90 92 94 96 98 100
Eve
nts
/ 2.5
GeV
2
4
6
8
10
12 ATLAS-1 = 8 TeV, 20.3 fbs
µµSR-Z
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
[GeV]missTE
200 250 300 350 400 450 500
Eve
nts
/ 25
GeV
2
4
6
8
10
12ATLAS
-1 = 8 TeV, 20.3 fbs
µµSR-Z
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
11 / 35 Alex Martyniuk BSM searches at ATLAS
Summary – ll + jets + EmissT
Channel searches for SUSY cascade decays with SF leptons, jets and EmissT both
on and off Z mass shellCombined significance 3.0σExcesses seen in both on shell channels, therefore weaker limitsDevil’s advocate/pessimist in me:
≈ 180 SR in SUSY searches Run-1Would naively expect ≈ 0.5× 3σ excesses and have seen oneOnly Run-2 will tell
[GeV]llm82 84 86 88 90 92 94 96 98 100
Eve
nts
/ 2.5
GeV
2
4
6
8
10
12
14ATLAS
-1 = 8 TeV, 20.3 fbs
SR-Z ee
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
[GeV]µ200 400 600 800 1000
) [G
eV]
g~m
(
600
700
800
900
1000
1100
1200
1300
1400
)0
1χ∼
) < m
(
g~m(
200 300 400 500 600 700 ) [GeV]0
1χ∼m(
)=1.5 TeVq~ = 1 TeV, m(2 = M1
= 1.5, Mβ; tan0
1χ∼GGM: higgsino-like
-1=8 TeV, 20.3 fbs
µµSR-Z ee+
ATLAS )theorySUSYσ1 ±Observed limit (
)expσ1 ±Expected limit (
)expσ2 ±Expected limit (
12 / 35 Alex Martyniuk BSM searches at ATLAS
Exotic Searches – LLP
Model Signature∫L dt[fb−1] Lifetime limit Reference
SU
SY
Hig
gsB
R=
10%
Hig
gsB
R=
5%30
0G
eVsc
alar
900
GeV
scal
arO
ther
RPV χ01 → eeν/eµν/µµν displaced lepton pair 20.3 1504.05162m(g̃) = 1.3 TeV, m(χ01) = 1.0 TeV7-740 mmχ01
lifetime
GGM χ01 → ZG̃ displaced vtx + jets 20.3 1504.05162m(g̃) = 1.1 TeV, m(χ01) = 1.0 TeV6-480 mmχ01
lifetime
AMSB pp → χ±1χ01,χ+1 χ−1 disappearing track 20.3 1310.3675m(χ±1 ) = 450 GeV0.22-3.0 mχ±1
lifetime
AMSB pp → χ±1χ01,χ+1 χ−1 large pixel dE/dx 18.4 1506.05332m(χ±1 ) = 450 GeV1.31-9.0 mχ±1
lifetime
GMSB non-pointing or delayed γ 20.3 1409.5542SPS8 with Λ = 200 TeV0.08-5.4 mχ01
lifetime
Stealth SUSY 2 ID/MS vertices 19.5 1504.03634m(g̃) = 500 GeV0.12-90.6 mS̃ lifetime
Hidden Valley H → πvπv 2 low-EMF trackless jets 20.3 1501.04020m(πv) = 25 GeV0.41-7.57 mπv lifetime
Hidden Valley H → πvπv 2 ID/MS vertices 19.5 1504.03634m(πv) = 25 GeV0.31-25.4 mπv lifetime
FRVZ H → 2γd + X 2 e−, µ−,π−jets 20.3 1409.0746H → 2γd + X , m(γd ) = 400 MeV14-140 mmγd lifetime
FRVZ H → 4γd + X 2 e−, µ−,π−jets 20.3 1409.0746H → 4γd + X , m(γd ) = 400 MeV15-260 mmγd lifetime
Hidden Valley H → πvπv 2 low-EMF trackless jets 20.3 1501.04020m(πv) = 25 GeV0.6-5.0 mπv lifetime
Hidden Valley H → πvπv 2 ID/MS vertices 19.5 1504.03634m(πv) = 25 GeV0.43-18.1 mπv lifetime
FRVZ H → 4γd + X 2 e−, µ−,π−jets 20.3 1409.0746H → 4γd + X , m(γd ) = 400 MeV28-160 mmγd lifetime
Hidden Valley Φ→ πvπv 2 low-EMF trackless jets 20.3 1501.04020σ×BR = 1 pb, m(πv) = 50 GeV0.29-7.9 mπv lifetime
Hidden Valley Φ→ πvπv 2 ID/MS vertices 19.5 1504.03634σ×BR = 1 pb, m(πv) = 50 GeV0.19-31.9 mπv lifetime
Hidden Valley Φ→ πvπv 2 low-EMF trackless jets 20.3 1501.04020σ×BR = 1 pb, m(πv) = 50 GeV0.15-4.1 mπv lifetime
Hidden Valley Φ→ πvπv 2 ID/MS vertices 19.5 1504.03634σ×BR = 1 pb, m(πv) = 50 GeV0.11-18.3 mπv lifetime
HV Z ′(1 TeV)→ qvqv 2 ID/MS vertices 20.3 1504.03634σ×BR = 1 pb, m(πv) = 50 GeV0.1-4.9 mπv lifetime
HV Z ′(2 TeV)→ qvqv 2 ID/MS vertices 20.3 1504.03634σ×BR = 1 pb, m(πv) = 50 GeV0.1-10.1 mπv lifetime
cτ [m]0.01 0.1 1 10 100√s = 8 TeV
ATLAS Long-lived Particle Searches* - 95% CL ExclusionStatus: July 2015
ATLAS Preliminary∫L dt = (18.4 - 20.3) fb−1
√s = 8 TeV
*Only a selection of the available lifetime limits on new states is shown.
13 / 35 Alex Martyniuk BSM searches at ATLAS
Exotic Searches – needles in the haystack?
14 / 35 Alex Martyniuk BSM searches at ATLAS
Exotics search in ll + b − jets + EmissT
[EXOT-2013-16]g
T̄
Z
g
g
T
H
t
t̄
g
σ
t
t̄
t
g
g
σ
t̄
u, c
u, c t
H
t
Look for excess of events with same-sign (SS) lepton pairs, b-jets and EmissT
Using a simple ‘cut and count’ methodology
As most Exotics analyses do, the channel is sensitive to a wide range of BSMtheories
Vector-like quarks (VLQ)Any enhancement of t t̄ t t̄ production (g̃ pair production, 2UED/RPP)Chiral quark production, b′ → WtSame-sign tt production via flavour changing neutral currents (FCNC)
15 / 35 Alex Martyniuk BSM searches at ATLAS
Signal regionsg
σ
t
t̄
t
g
g
σ
t̄
Backgrounds studiedReal SS lepton pairs W ±W ± jj , t t̄W/Z ,t t̄W +W−, t t̄H, WH, ZH, tWZ , tH, WZ , ZZ andt t̄ t t̄ : MC modelled (low x-sections)Fake leptons: Data-driven. Derived from matrixmethod.Charge misidentification: Rate estimated inZ → ee events
Estimates validated in low-HT region,100 > HT > 400 GeV
Definition Namee±e± + e±µ± + µ±µ± + eee+ eeµ+ eµµ+ µµµ, Nj ≥ 2
400 < HT < 700GeVNb = 1 SRVLQ0Nb = 2 Emiss
T > 40 GeV SRVLQ1 SR4t0Nb ≥ 3 SRVLQ2 SR4t1
HT ≥ 700 GeV
Nb = 140 < Emiss
T < 100GeV SRVLQ3Emiss
T ≥ 100 GeV SRVLQ4
Nb = 240 < Emiss
T < 100GeV SRVLQ5 SR4t2Emiss
T ≥ 100 GeV SRVLQ6 SR4t3Nb ≥ 3 Emiss
T > 40 GeV SRVLQ7 SR4t4e+e+, e+µ+, µ+µ+, Nj ∈ [2, 4], ∆φ`` > 2.5
HT > 450 GeV Nb ≥ 1 EmissT > 40 GeV SRttee, SRtteµ, SRttµµ
16 / 35 Alex Martyniuk BSM searches at ATLAS
Preselection Results
Preselection appliedEssentiallySRLQ0/1/2 withoutthe HT or Emiss
Tselections applied
Good data/backgroundagreement withinuncertainties seen inpreselection withone/two b-tagged jets
Good agreement alsoseen in the validationregions
[GeV]missTE
0 50 100 150 200 250 300 E
vent
s / 1
0 G
eV0
20
40
60
80
100
120
140
160
180
200 Data
W/Ztt
Q Mis-id
Non-prompt/fakes
Htt
Dibosons
Other
Uncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
preselection + 1 b-jet
[GeV]missTE
0 50 100 150 200 250 300
Eve
nts
/ 10
GeV
0
10
20
30
40
50
60
Data
W/Ztt
Q Mis-id
Non-prompt/fakes
Htt
Dibosons
Other
Uncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
preselection + 2 b-jets
[GeV]TH0 200 400 600 800 1000 1200 1400 1600
Eve
nts
/ 50
GeV
0
50
100
150
200
250
300
Data
W/Ztt
Q Mis-id
Non-prompt/fakes
Htt
Dibosons
Other
Uncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
preselection + 1 b-jet
[GeV]TH0 200 400 600 800 1000 1200 1400 1600
Eve
nts
/ 50
GeV
0
10
20
30
40
50
60
70
80
90Data
W/Ztt
Q Mis-id
Non-prompt/fakes
Htt
Dibosons
Other
Uncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
preselection + 2 b-jets
17 / 35 Alex Martyniuk BSM searches at ATLAS
SR Results
Excess of 2.0σ seen taking all SRVLQs into account (VLQ and b′ models)Excess of 2.5σ seen taking only 4-top SRs into accountSS+FCNC SRs are in good agreement with SM expectation
Eve
nts
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
22000
24000Data
W/ZttQ Mis-idFake/non-prompt leptons
HttDibosonsOtherUncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
SRVLQ0 SR4t0SRVLQ1
SR4t1SRVLQ2 SRVLQ3 SRVLQ4
SR4t2SRVLQ5
SR4t3SRVLQ6
SR4t4SRVLQ7
Sig
nific
ance
-3-2-10123
Eve
nts
-110
1
10
210
310
410Data
W/ZttQ Mis-idFake/non-prompt leptons
HttDibosonsOtherUncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
SRVLQ0 SR4t0SRVLQ1
SR4t1SRVLQ2 SRVLQ3 SRVLQ4
SR4t2SRVLQ5
SR4t3SRVLQ6
SR4t4SRVLQ7
Sig
nific
ance
-3-2-10123
18 / 35 Alex Martyniuk BSM searches at ATLAS
Summary – ll + b − jets + EmissT
Cut/count analysis looking for events with SS lepton pairs and b-jets as a signalfor many BSM signalsExcesses of 2.5σ(2.0σ) seen taking 4-top SRs(all SRs) into accountMass resolution of the analysis is poorLimits worse than expected due to presence of excessAgain Run-2 is needed to to elucidate the origin of this excess
Signal mass [TeV]
0.4 0.6 0.8 1 1.2
p-va
lue
-310
-210
-110
1
σ1
σ2
σ3
ATLAS-1 = 8 TeV, 20.3 fbs
ATLAS-1 = 8 TeV, 20.3 fbs
SgluonUED-RPP
Sgluon mass [TeV]
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
) [p
b]ttt
BR
(t× σ
-310
-210
-110
1
10
ATLAS-1 = 8 TeV, 20.3 fbs
95% CL observed limit
95% CL expected limit
95% CL expected limitσ 1±
95% CL expected limitσ 2±
) = 100%]t t→ σTheory [BR(
19 / 35 Alex Martyniuk BSM searches at ATLAS
Exotics search in X → VV → JJ
[EXOT-2013-08] Bias starts here
Diboson resonances appear in manyextensions to the standard model
The following analysis concentrates ontwo benchmark models
Extended gauge sector models(W ′ → WZ )Extra dimensions models(GRS → WW/ZZ )
Low branching ratios hinder theleptonic searches at the highest masses
Obviously, a fully hadronic search hasaccess to these lost events
The problem, is controlling theenormous QCD background that theleptonic searches were avoiding
W ⇓ Diboson branching ratioslν (33%) 23% 7% 3%qq (67%) 47% 13% 7%
Z ⇒ qq (70%) νν (20%) ll (10%)
Use large-R jets and jet substructureto select signal events from dominantQCD multi-jet backgrounds
20 / 35 Alex Martyniuk BSM searches at ATLAS
Boosted Bosons
WI WZ Large ΔR
Large ΔR
Low Mass <500GeV
WI WZ
High Mass >1TeV
Small ΔR
Small ΔR
Increasing Mass
Vector bosons have massO(0.1 TeV)
We are interested in particlesof mass ≥ O(1 TeV)
Therefore the decays of theform, X → VV with large mX ,lead to vector bosons withvery high pT
Therefore, boosted decayproducts become morecollimatedRule of thumb for angularseparation of decay products:
∆R =√
∆φ2 + ∆η2 ≈ 2mpT
21 / 35 Alex Martyniuk BSM searches at ATLAS
Fat Jet→Grooming→Tagging1 Reconstruct decay as fat-jet
Use large-R parameter jet to collect radiation from the original decay2 Groom the jet
Signal: Remove unwanted jet constituents not from the signal, e.g. pile-upBackground: Preserve the background characteristics
3 Tag as boson jetUse differences between signal and background jet characteristics to rejectbackground jets
22 / 35 Alex Martyniuk BSM searches at ATLAS
BDRS-A Split-Filter
In this analysis a modified BDRS-Asplit filtering algorithm is used
Starts from R = 1.2 CA jets seededfrom local cluster weighted (LCW)topological clusters
Loose BDRS tagger, with no massdrop requirement
Iterative parameter Value√ymin 0.20µmax 1.00
Filtering parameter Valuenr 3Rr 0.3
23 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
10
210
310
410
510
610
710
810
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
No boson tagging
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
|jj
y∆|0 0.5 1 1.5 2 2.5 3
Eve
nts
/ 0.1
0
20
40
60
80
100
120
140
160
310×
W' (1400 GeV) x 3000QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
< 1.54 TeVjj m≤1.26
|jj
y∆|0 0.5 1 1.5 2 2.5 3
Eve
nts
/ 0.1
0
2000
4000
6000
8000
10000
12000
14000 W' (2400 GeV) x 15000QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
< 2.64 TeVjj m≤2.16
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
)T2
+pT1
)/(pT2
-pT1
(p0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Eve
nts
/ 0.0
1
0
10000
20000
30000
40000
50000W' (1400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆|| < 2η|
< 1.54 TeVjj m≤1.26
)T2
+pT1
)/(pT2
-pT1
(p0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Eve
nts
/ 0.0
1
0
200
400
600
800
1000
1200
1400 W' (2400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆|| < 2η|
< 2.64 TeVjj m≤2.16
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
ηjet -3 -2 -1 0 1 2 3
Jets
/ 0.
1
0
10000
20000
30000
40000
50000W' (1400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆| < 1.54 TeVjj m≤1.26
ηjet -3 -2 -1 0 1 2 3
Jets
/ 0.
1
0
200
400
600
800
1000
1200
1400
1600 W' (2400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆| < 2.64 TeVjj m≤2.16
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
ηjet -3 -2 -1 0 1 2 3
Jets
/ 0.
1
0
10000
20000
30000
40000
50000W' (1400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆| < 1.54 TeVjj m≤1.26
ηjet -3 -2 -1 0 1 2 3
Jets
/ 0.
1
0
200
400
600
800
1000
1200
1400
1600 W' (2400 GeV) x 2500QCD (Pythia)QCD (Herwig)QCD (PowPy)Data
ATLAS = 8 TeVs -1 , 20.3fb
| < 1.2jj
y∆| < 2.64 TeVjj m≤2.16
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
[TeV]jm0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Fra
ctio
n of
jets
/ 5.
0 G
eV
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22 ATLAS Simulation
= 8 TeVs
= 1.8 TeV)G
WW (m→ RS
bulk G
= 1.8 TeV)G
ZZ (m→ RS
bulk G
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
> 0.45y
yJet 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fra
ctio
n of
jets
/ 0.
05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35ATLAS Simulation
= 8 TeVs = 1.8 TeV)
W' WZ (m→EGM W'
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
< 110 GeVj m≤60
trkJet ungroomed n
0 20 40 60 80 100 120
Fra
ctio
n of
jets
/ 3
0
0.05
0.1
0.15
0.2
0.25ATLAS Simulation
= 8 TeVs = 1.8 TeV)
W' WZ (m→EGM W'
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
< 110 GeVj m≤60 > 0.45 y
24 / 35 Alex Martyniuk BSM searches at ATLAS
Event selection1 Trigger: pT > 360 GeV anti-kt 1.02 Apply BDRS-A split-filter3 Require mJJ > 1.05 TeV
Ensures on trigger plateau
4 Rapidity gap between leading jets,|∆y12| < 1.2
s-channel signal more central thant-channel QCD
5 Leading jets pT asymmetry ApT < 0.15Used as proxy for large-R jet cleaning
6 Leading jets |η| < 2.0Ensures a good overlap with tracker
7 Correction for jets on calorimetry holes8 Boson tagging cuts
Jet mass (WZ , WW , ZZ ), momentumbalance, ntrk ghost associated to jet
Background efficienciesTopological ε ≈ 48%Tagger ε ≈ 1.2− 0.6% (per jet)
Resonance Mass [TeV]1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Sel
ectio
n ef
ficie
ncy
0.5
0.6
0.7
0.8
0.9
1
1.1ATLAS Simulation
= 8 TeVs WZ→EGM W'
WW→ RSbulk G
ZZ→ RSbulk G
Event topology requirements
Resonance Mass [TeV]1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Sel
ectio
n ef
ficie
ncy
0
0.05
0.1
0.15
0.2
0.25
0.3ATLAS Simulation
= 8 TeVs WZ→EGM W'
WW→ RSbulk G
ZZ→ RSbulk G
24 / 35 Alex Martyniuk BSM searches at ATLAS
X →WZ selection
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
Full WZ selection appliedZ mass window appliedto leading mass jetW mass windowapplied to sub-leadingmass jet
Good agreement seen withsteeply, smoothly fallingbackground model in thelow/high mass regions
Deviation from thebackground observed ataround 2 TeV
25 / 35 Alex Martyniuk BSM searches at ATLAS
X →WW selection
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WW Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
Full WW selection applied tothe data
W mass windowapplied to both jets
Good agreement again seenwith steeply, smoothly fallingbackground model
Deviation from thebackground still observed ataround 2 TeV
Remember: There is anoverlap between the W /Zmass windows
26 / 35 Alex Martyniuk BSM searches at ATLAS
X → ZZ selection
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
ZZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
Full ZZ selection applied tothe data
Z mass windowapplied to both jets
Good agreement again seenwith steeply, smoothly fallingbackground model
Deviation from thebackground still observed ataround 2 TeV
Remember: There is anoverlap between the W /Zmass windows
27 / 35 Alex Martyniuk BSM searches at ATLAS
Observed p0 value
[GeV]jjm1400 1600 1800 2000 2200 2400 2600 2800 3000
0Lo
cal p
4−10
3−10
2−10
1−10
1
10WZ Selection
WW Selection
ZZ Selection
ATLAS-1=8TeV, 20.3 fbs
σ0
σ1
σ2
σ3
σ3.5
σ4
Therefore, no statistically significant deviation fromthe background has been observed
A discrepancy was seenwith respect to theexpected backgrounddistribution
Once suitably confident itis not an error, itssignificance should bequantifiedIn the WZ channel:
Local p0 = 3.4σ
Global p0 = 2.5σGlobal σ takes intoaccount the lookelsewhere effectLEE includes weightedcontribution fromWW /ZZ channels dueto the overlap
28 / 35 Alex Martyniuk BSM searches at ATLAS
X →WZ limits
As no significant deviation was observed, we continue to set limits on theobserved distributions
95% confidence limits set on σ × B using the CLS prescription taking intoaccount the systematic uncertainties and background fit
[TeV]W' m1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
WZ
) [fb
]→
BR
(W'
× W
') →
(pp
σ
1−10
1
10
210
310
410 Observed 95% CL
Expected 95% CL
uncertaintyσ 1±
unceirtaintyσ 2±
EGM W', c = 1
ATLAS-1 = 8 TeV, 20.3 fbs
Expected limits broadlyagree with the observedlimits
Exclusion of EGM W ′
from 1.3− 1.5 TeV
Broad deviation from thebackground observable ataround 2 TeV
Benchmark extendedgauge model W ′ σ × Bshown for comparisonpurposes
WW , ZZ in backup
29 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? ATLAS combination
ATLAS diboson combination ATLAS-CONF-2015-045
In the full combination discrepancy is reduced to a 2.5σ local excess
Limits increased to 1.81TeV in the W ′ search channels
Fully hadronic channel in (significant) tension with the leptonic channelsPointing to statistical fluctuation in fully hadronic channel?Run-2 will sort this out!
Full paper following up CONF note in the ATLAS pipeline
30 / 35 Alex Martyniuk BSM searches at ATLAS
Summary – X → VV → JJ
Presented a search for a high mass dibosonresonance, [arXiv:1506.00962]
Used 20.3fb−1 8TeV ATLAS dataJet substructure (BDRS-A CA 1.2 jets) used toseparate signal from backgroundQCD dominated background modelled byparametric function
Deviation from expected steeply, smoothly fallingbackground seen at 2 TeVCross-checks performed, no major issuesdiscovered
Excess p0 = 3.4σ local, 2.5σ global
Limits exclude EGM W ′ models with1.3 < mW ′ < 1.5 TeV
Combination with other diboson search channelsreduces excess to 2.5σ local
Preparations underway to repeat search in13TeV data
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
[TeV]W' m1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
WZ
) [fb
]→
BR
(W'
× W
') →
(pp
σ
1−10
1
10
210
310
410 Observed 95% CL
Expected 95% CL
uncertaintyσ 1±
unceirtaintyσ 2±
EGM W', c = 1
ATLAS-1 = 8 TeV, 20.3 fbs
31 / 35 Alex Martyniuk BSM searches at ATLAS
The Future: Run-2
32 / 35 Alex Martyniuk BSM searches at ATLAS
The Future How soon is now?
33 / 35 Alex Martyniuk BSM searches at ATLAS
How much data is needed?
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
So question one is how muchRun-2 13 TeV data do we needto surpass for example theRun-1 VV → JJ result?
At 2 TeV productioncross-sections growconsiderably
For gluon-gluon fusion ×15,i.e. for GRS productionFor qq̄ initiated ×8, i.e. forW′ productionUnfortunately QCDbackground also increases
So a smaller amount of Run-2data (1− 3fb−1) should beroughly equivalent to the 8 TeVdataset
34 / 35 Alex Martyniuk BSM searches at ATLAS
How much data is needed?
100 10001
10
100
gg
Σqq
qg
WJS2013
ratios of LHC parton luminosities: 13 TeV / 8 TeV
lum
ino
sity r
atio
MX (GeV)
MSTW2008NLO
_
So question one is how muchRun-2 13 TeV data do we needto surpass for example theRun-1 VV → JJ result?
At 2 TeV productioncross-sections growconsiderably
For gluon-gluon fusion ×15,i.e. for GRS productionFor qq̄ initiated ×8, i.e. forW′ productionUnfortunately QCDbackground also increases
So a smaller amount of Run-2data (1− 3fb−1) should beroughly equivalent to the 8 TeVdataset
34 / 35 Alex Martyniuk BSM searches at ATLAS
How much data is needed?
100 10001
10
100
gg
Σqq
qg
WJS2013
ratios of LHC parton luminosities: 13 TeV / 8 TeV
lum
ino
sity r
atio
MX (GeV)
MSTW2008NLO
_
So question one is how muchRun-2 13 TeV data do we needto surpass for example theRun-1 VV → JJ result?
At 2 TeV productioncross-sections growconsiderably
For gluon-gluon fusion ×15,i.e. for GRS productionFor qq̄ initiated ×8, i.e. forW′ productionUnfortunately QCDbackground also increases
So a smaller amount of Run-2data (1− 3fb−1) should beroughly equivalent to the 8 TeVdataset
34 / 35 Alex Martyniuk BSM searches at ATLAS
Conclusions
Presented a series of searches for BSM physics by the ATLAS SUSY andExotics groups in Run-1 of the LHC which showed anomalous features
All used 20.3fb−1 8TeV ATLAS data
Deviations between 2− 3.5σ local observed
In the VV → JJ search a global significance of 2.5σ was observed, beforecombination with the leptonic channels
[GeV]llm82 84 86 88 90 92 94 96 98 100
Eve
nts
/ 2.5
GeV
2
4
6
8
10
12
14ATLAS
-1 = 8 TeV, 20.3 fbs
SR-Z ee
DataStandard ModelFlavour SymmetricOther Backgrounds
=(700,200)GeVµ),g~m(=(900,600)GeVµ),g~m(
Eve
nts
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
22000
24000Data
W/ZttQ Mis-idFake/non-prompt leptons
HttDibosonsOtherUncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
SRVLQ0 SR4t0SRVLQ1
SR4t1SRVLQ2 SRVLQ3 SRVLQ4
SR4t2SRVLQ5
SR4t3SRVLQ6
SR4t4SRVLQ7
Sig
nific
ance
-3-2-10123
Eve
nts
-110
1
10
210
310
410Data
W/ZttQ Mis-idFake/non-prompt leptons
HttDibosonsOtherUncertainties
ATLAS-1 = 8 TeV, 20.3 fbs
SRVLQ0 SR4t0SRVLQ1
SR4t1SRVLQ2 SRVLQ3 SRVLQ4
SR4t2SRVLQ5
SR4t3SRVLQ6
SR4t4SRVLQ7
Sig
nific
ance
-3-2-10123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
35 / 35 Alex Martyniuk BSM searches at ATLAS
Backup
Backup
35 / 35 Alex Martyniuk BSM searches at ATLAS
SUSY: Z/γ+jets modellingE
vent
s / 1
0 G
eV
1
10
210
310
410
DataStandard ModelZ+jets (Jet Smearing)Flavour SymmetricOther Backgrounds
-1 = 8 TeV, 20.3 fbs
VRZ ee
) > 0.4miss
T,E
1,2 (jetφ ∆
ATLAS
[GeV]missTE
0 10 20 30 40 50 60 70 80 90 100
Dat
a / S
M
00.5
11.5
2
Eve
nts
/ 10
GeV
1
10
210
310
410
DataStandard ModelZ+jets (Jet Smearing)Flavour SymmetricOther Backgrounds
-1 = 8 TeV, 20.3 fbs
µµVRZ
) > 0.4miss
T,E
1,2 (jetφ ∆
ATLAS
[GeV]missTE
0 10 20 30 40 50 60 70 80 90 100
Dat
a / S
M
00.5
11.5
2
35 / 35 Alex Martyniuk BSM searches at ATLAS
Cambridge-Aachen Jets
Cambridge-Aachen jets (CA jets)[arXiv:9707323] or [arXiv:0802.2470]
Part of the sequential recombinationfamily of jet reconstruction algorithms
Calculate the ∆Rij between all jetconstituentsCombine closest constituents firstMerge while R ≤ 1.2 (in this analysis)If there are no components within 1.2,redefine as a jet and remove from thecollection of constituentsMerge until there are no componentsleft
NO pT dependence!
Therefore can look into the history anduse the pT splitting information
35 / 35 Alex Martyniuk BSM searches at ATLAS
BDRS Split-FilterThe BDRS split filtering algorithm, [arXiv:0802.2470], decomposes CA jetssequential clustering to find hard substructure withinOriginally defined to find boosted H → bb decays
The decomposition follows some simple stepsFor jet j , undo the last step of clustering forming jets j1 and j2 (mj1 > mj2 )If there was a large mass drop, mj1 < µmaxmj and the pT balance is not too
asymmetric,min(p2
Tj1,p2
Tj2)
m2j0
∆R2j1,j2≥ ymin, define j as from a hard splitting and stop
Otherwise redefine j as j1, discard j2, and continueFilter the resulting jet by re-clustering as nr × Rr sized subjets
35 / 35 Alex Martyniuk BSM searches at ATLAS
BDRS-A CA R = 1.2 Jet Calibration
Particle level jet energy and mass calibrations were derived and applied to theBDRS-A CA R = 1.2 jets used in the analysisEffectively restores jet energy/mass response over the full jet E and η range
Calculate the jet energyresponse in bins of ηdetand Etruth
Fit the responses with aGaussian fit, to gainmean response in eachbin, < Rjet
E >
Derive the meanreconstructed jet energy,< Ejet
reco >
Fit the < RjetE > vs
< Ejetreco > distribution to
gain a calibrationfunctionRepeat process formass calibration usingthe LCW+JES jets
ηJet -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Jet e
nerg
y re
spon
se b
efor
e ca
libra
tion
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2E = 450 GeVE = 800 GeVE = 1100 GeVE = 1400 GeVE = 1800 GeV
ATLAS Simulation
BDRS-A jets, C/A R = 1.2
Tagging selection applied
= 8 TeVsPythia 8 dijets,
ηJet -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Jet e
nerg
y re
spon
se a
fter
calib
ratio
n
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2E = 450 GeVE = 800 GeVE = 1100 GeVE = 1400 GeVE = 1800 GeV
ATLAS Simulation
BDRS-A jets, C/A R = 1.2
Tagging selection applied
= 8 TeVsPythia 8 dijets,
ηJet -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Jet m
ass
resp
onse
bef
ore
calib
ratio
n
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2E = 450 GeVE = 800 GeVE = 1100 GeVE = 1400 GeVE = 1800 GeV
ATLAS Simulation
BDRS-A jets, C/A R = 1.2
Tagging selection applied
= 8 TeVsPythia 8 dijets,
ηJet -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Jet m
ass
resp
onse
afte
r ca
libra
tion
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2E = 450 GeVE = 800 GeVE = 1100 GeVE = 1400 GeVE = 1800 GeV
ATLAS Simulation
BDRS-A jets, C/A R = 1.2
Tagging selection applied
= 8 TeVsPythia 8 dijets,
35 / 35 Alex Martyniuk BSM searches at ATLAS
Tools: Jet mass
2 Filtered jet massSeparates peaked boson mass from falling QCD spectrum
Apply ±13 GeV window cutsaround boson mass from MCsimulation peak (mW = 82.4,mZ = 92.8)
For example, in the WZ cut;Leading mass jet79.8 GeV < mjet < 105.8 GeVSubleading mass jet69.4 GeV < mjet < 95.4 GeV
Very powerful cut!εsignal ≈ 80%εbackground ≈ 10− 15%
Cuts optimised using a data CRDijet formed from twotagged/un-tagged regions
N.B. Windows overlap!!!
[TeV]jm0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Fra
ctio
n of
jets
/ 5.
0 G
eV
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22 ATLAS Simulation
= 8 TeVs
= 1.8 TeV)G
WW (m→ RS
bulk G
= 1.8 TeV)G
ZZ (m→ RS
bulk G
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
> 0.45y
35 / 35 Alex Martyniuk BSM searches at ATLAS
Tools: Subjet momentum balance
3 Subjet momentum balanceBoson jets symmetric, QCD unbalanced
Soft gluon radiation leads toasymmetric splittings
W/Z → qq̄ decays tend to sharemomentum more equally betweendecay products
Apply a more stringent√
y ≥ 0.45cut on the subjet momentumbalance
Another powerful cut!εsignal ≈ 70%εbackground ≈ 30%
Cuts optimised using MC, using awide mass window,60 GeV < mjet < 110 GeV
yJet 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fra
ctio
n of
jets
/ 0.
05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35ATLAS Simulation
= 8 TeVs = 1.8 TeV)
W' WZ (m→EGM W'
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
< 110 GeVj m≤60
35 / 35 Alex Martyniuk BSM searches at ATLAS
Tools: Ghost matched jets4 Number of tracks ghost matched to the unfiltered jet
More hadronic activity in QCD jets
Emission of hard gluon dominatesafter mass/asymmetry cuts
Expect increased hadronic activityfrom gluon
Use the number of ghostassociated ungroomed tracks, ntrk,as a proxy for hadronic activity,[arXiv:0802.1188]
Apply ntrk ≤ 30 cut
Efficiency after mass/asymmetryεsignal = 83± 7%εbackground ≈ 65%
Very hard to model in MC
Cuts optimised using V+jetsenriched data CREfficiency calibrated in this CR
trkJet ungroomed n
0 20 40 60 80 100 120
Fra
ctio
n of
jets
/ 3
0
0.05
0.1
0.15
0.2
0.25ATLAS Simulation
= 8 TeVs = 1.8 TeV)
W' WZ (m→EGM W'
Pythia QCD dijet
| < 1.2jj
y∆|| < 2
jη|
< 1.98 TeVjj m≤1.62
< 110 GeVj m≤60 > 0.45 y
35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
10
210
310
410
510
610
710
810
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
No boson tagging
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Where,x = mJJ/
√s
mJJ is the dijet invariant mass,p1 is a normalisation factor,p2 and p3 are dimensionless shapeparametersξ is a dimensionless constant chosen afterfitting to minimise the correlations betweenp2 and p3
35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
10
210
310
410
510
610
710
810
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
No boson tagging
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Where,x = mJJ/
√s
mJJ is the dijet invariant mass,p1 is a normalisation factor,p2 and p3 are dimensionless shapeparametersξ is a dimensionless constant chosen afterfitting to minimise the correlations betweenp2 and p3
35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
10
210
310
410
510
610
710
810
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
No boson tagging
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Sample χ2/nDOF ProbabilityPythia dijet events 24.6/22 0.31Herwig++ dijet events 15.9/22 0.82Data with 110 < mj1 ≤ 140 GeV and 40 < mj2 ≤ 60 GeV 12.1/11 0.79Data with 40 < mj ≤ 60 GeV for both jets 19.8/13 0.56Data with 110 < mj ≤ 140 GeV for both jets 5.0/6 0.91
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Error taken from errors on functionalparameters
Fit tested on,Raw dataPYTHIA/HERWIG MCMass sideband data CRs
Alternate fit functions give similar results35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
< 60 GeVj
40 < m
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Sample χ2/nDOF ProbabilityPythia dijet events 24.6/22 0.31Herwig++ dijet events 15.9/22 0.82Data with 110 < mj1 ≤ 140 GeV and 40 < mj2 ≤ 60 GeV 12.1/11 0.79Data with 40 < mj ≤ 60 GeV for both jets 19.8/13 0.56Data with 110 < mj ≤ 140 GeV for both jets 5.0/6 0.91
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Error taken from errors on functionalparameters
Fit tested on,Raw dataPYTHIA/HERWIG MCMass sideband data CRs
Alternate fit functions give similar results35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
< 60 GeVj1,j2
40 < m
< 140 GeVj2,j1
110 < m
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Sample χ2/nDOF ProbabilityPythia dijet events 24.6/22 0.31Herwig++ dijet events 15.9/22 0.82Data with 110 < mj1 ≤ 140 GeV and 40 < mj2 ≤ 60 GeV 12.1/11 0.79Data with 40 < mj ≤ 60 GeV for both jets 19.8/13 0.56Data with 110 < mj ≤ 140 GeV for both jets 5.0/6 0.91
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Error taken from errors on functionalparameters
Fit tested on,Raw dataPYTHIA/HERWIG MCMass sideband data CRs
Alternate fit functions give similar results35 / 35 Alex Martyniuk BSM searches at ATLAS
Modelling the Background
After trying to kill the background we now arrive at the point of modelling itMC statistics needed to properly model the high mJJ tail are prohibitively large
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
< 140 GeVj
110 < m
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
Sample χ2/nDOF ProbabilityPythia dijet events 24.6/22 0.31Herwig++ dijet events 15.9/22 0.82Data with 110 < mj1 ≤ 140 GeV and 40 < mj2 ≤ 60 GeV 12.1/11 0.79Data with 40 < mj ≤ 60 GeV for both jets 19.8/13 0.56Data with 110 < mj ≤ 140 GeV for both jets 5.0/6 0.91
Assume a steeply and smoothly fallingdistribution models the background
Any resonance should be narrow, thus onlyaffect a few bins
Use a parametric function to model thebackground from the data
dndx
= p1(1− x)p2−ξp3 xp3
Error taken from errors on functionalparameters
Fit tested on,Raw dataPYTHIA/HERWIG MCMass sideband data CRs
Alternate fit functions give similar results35 / 35 Alex Martyniuk BSM searches at ATLAS
Systematic uncertainties: Shape
Background: Taken from the uncertainties on the fit parametersSignal: Various systematics affect the signal reconstruction and selectionefficiency
Shape systematics:The jet pT and jet mass scaleuncertainties determined by thetrack/calo double ratio technique
For example for a variable x,
xdatatrack/x
datacalo
xMCtrack/x
MCcalo
Applied as a Gaussian with µ = 1 and σequal to the observed uncertainty
jet pT scale: 2%jet mass scale: 3%
An uncertainty on the jet pT resolution of20% is applied as an additionalsmearing on top of the nominal 5%
) [GeV]2
, m1
max(m0 20 40 60 80 100 120 140 160
Jets
/ 5
GeV
0
2
4
6
8
10
12
14
16
18 ATLAS Simulation = 8 TeVs -1 , 20.3fb
BDRS-A| < 1.2
jjy∆|| < 2.0η|
A < 0.15 >= 0.45
fy
< 30trkn
qqqq→ WZ →W'(1.4 TeV)
NominalJMS upJMS downJMR
Source Uncertainty Constraining pdfJet pT scale 2% G(αPT|1, 0.02)
Jet pT resolution 20% G(σrE | 0, 0.05×√1.22 − 12)
Jet mass scale 3% G(αm|1, 0.03)
35 / 35 Alex Martyniuk BSM searches at ATLAS
Systematic uncertainties: Normalisation
Background: Taken from the uncertainties on the fit parametersSignal: Various systematics affect the signal reconstruction and selectionefficiency
Normalisation systematics:Large uncertainty on the ntrk cutevaluated in the data driven V+jetsstudy used to define the efficiency of thecut
Jet mass scale affects both shape andnormalisation strongly√
y scale evaluated using the doubleratio method
Resolutions taken as 20% smearingsShower model evaluated by comparingMC showered by PYTHIA or HERWIG
PDF4LHC method used to evaluatePDF uncertainties
ATLAS luminosity uncertaintyassumed
) [GeV]2
, m1
min(m0 20 40 60 80 100 120 140 160
Jets
/ 5
GeV
0
2
4
6
8
10
12
14
16
18
20ATLAS Simulation
= 8 TeVs -1 , 20.3fbBDRS-A
| < 1.2jj
y∆|| < 2.0η|
A < 0.15 >= 0.45
fy
< 30trkn
qqqq→ WZ →W'(1.4 TeV)
NominalJMS upJMS downJMR
Source UncertaintyEfficiency of the track-multiplicity cut 20.0%
Jet mass scale 5.0%Jet mass resolution 5.5%
Subjet momentum-balance scale 3.5%Subjet momentum-balance resolution 2.0%
Parton shower model 5.0%Parton distribution functions 3.5%
Luminosity 2.8%
35 / 35 Alex Martyniuk BSM searches at ATLAS
Selection overlapsComparing the effect of applying all signal selections at once, and the sameflavour selections at once
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WW+ZZ+WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
Data
Background model
Significance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WW+ZZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−024
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WW Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
ZZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
35 / 35 Alex Martyniuk BSM searches at ATLAS
What do these events look like? Dramatic!
35 / 35 Alex Martyniuk BSM searches at ATLAS
What do these events look like? Energetic!
35 / 35 Alex Martyniuk BSM searches at ATLAS
Digging deeper into the jets....These jet event displays take a bit more explanation, but offer a powerfulinsight into the analysis jets
The ATLAS detector volume isshown unfolded in y and φ
Inner detector track positions areshown as crosses
Black Tracks: From primaryvertexBlue Tracks: From secondaryvertices
Calorimeter deposits are displayedon the rainbow scale
The outlines of the CA 1.2 jets areshown
Black: Leading pT jetMauve: Sub-leading jet
Grey area: Sub-jets after filtering
y-4 -3 -2 -1 0 1 2 3 4
φ
0
1
2
3
4
5
6
-310
-210
-110
1
10
210
310
ATLAS = 8 TeVs
Run: 203027 Event: 5303984
35 / 35 Alex Martyniuk BSM searches at ATLAS
Digging deeper into the jets....
y-4 -3 -2 -1 0 1 2 3 4
φ
0
1
2
3
4
5
6
-310
-210
-110
1
10
210
310
ATLAS = 8 TeVs
Run: 201556 Event: 7295269
y-4 -3 -2 -1 0 1 2 3 4
φ
0
1
2
3
4
5
6
-310
-210
-110
1
10
210
310
ATLAS = 8 TeVs
Run: 201489 Event: 75855145
What do we see here?Subjets are highly collimatedPV tracks are highly correlated with the selected sub-jetsEnergy deposits concentrated in the sub-jetPile-up tracks/deposits sparsely distributed over the eventsSuccessfully picked the boson out of the pileup?
35 / 35 Alex Martyniuk BSM searches at ATLAS
Time to cross-check.....
Deviations from the expected background, especially ones at the tail of the datadistribution, mean one thing for physicists....
What on Earth did we do wrong?Try to evaluate any possible issues with the analysisOperation Cross-check beginsLook for mistakes, bugs or shaping effects in:
Detector/data taking effectsJet reconstruction effectsEvent selection effects
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
1−10
1
10
210
310
410DataBackground model1.5 TeV EGM W', c = 12.0 TeV EGM W', c = 12.5 TeV EGM W', c = 1Significance (stat)Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WW Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
3−10
2−10
1−10
1
10
210
310
410DataBackground model
= 1PIM, k/RS1.5 TeV Bulk G
= 1PIM, k/RS2.0 TeV Bulk GSignificance (stat)
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
ZZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
35 / 35 Alex Martyniuk BSM searches at ATLAS
Time to cross-check.....
From E. Kajomovitz35 / 35 Alex Martyniuk BSM searches at ATLAS
Time to cross-check.....
Started trawling throughSR/CR kinematic distributions,looking for unusual features inthe signal regions
Look at the effect of single cutson the distribution
Look at the effect of N − 1 cutson the distribution
Is one cut driving all of thedeviation?
Many, many, many..... more,you can only get so muchapproved...
[TeV]jjm1 1.5 2 2.5 3 3.5
ηS
ub-le
adin
g je
t
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Num
ber
of e
vent
s
0
5
10
15
20
25ATLAS
-1 = 8 TeV, 20.3 fbs
< 60 GeVj140 < m
< 60 GeVj240 < m
[TeV]jjm1 1.5 2 2.5 3 3.5
ηS
ub-le
adin
g je
t
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Num
ber
of e
vent
s
0
5
10
15
20
25
30
35ATLAS
-1 = 8 TeV, 20.3 fbs
< 105.8 GeVj167.5 < m
< 60 GeVj240 < m
[TeV]jjm1 1.5 2 2.5 3 3.5
ηS
ub-le
adin
g je
t
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Num
ber
of e
vent
s
0
5
10
15
20
25ATLAS
-1 = 8 TeV, 20.3 fbs
< 60 GeVj140 < m
< 105.8 GeVj267.5 < m
[TeV]jjm1 1.5 2 2.5 3 3.5
ηS
ub-le
adin
g je
t
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Num
ber
of e
vent
s
0
5
10
15
20
25
30
35
40
45ATLAS
-1 = 8 TeV, 20.3 fbs
< 105.8 GeVj167.5 < m
< 105.8 GeVj267.5 < m
35 / 35 Alex Martyniuk BSM searches at ATLAS
Time to cross-check.....
Started trawling throughSR/CR kinematic distributions,looking for unusual features inthe signal regions
Look at the effect of single cutson the distribution
Look at the effect of N − 1 cutson the distribution
Is one cut driving all of thedeviation?
Many, many, many..... more,you can only get so muchapproved...
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
trkonly jet ungroomed n
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
fyonly jet
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
jonly m
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
35 / 35 Alex Martyniuk BSM searches at ATLAS
Time to cross-check.....
Started trawling throughSR/CR kinematic distributions,looking for unusual features inthe signal regions
Look at the effect of single cutson the distribution
Look at the effect of N − 1 cutson the distribution
Is one cut driving all of thedeviation?
Many, many, many..... more,you can only get so muchapproved...
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
jw/o m
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
fyw/o jet
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
trkw/o jet ungroomed n
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123 1.5 2 2.5 3 3.5
Eve
nts
/ 100
GeV
2−10
1−10
1
10
210
310
410
510
610
710
810
Data
Background model
2.0 TeV EGM W', c = 1
Significance (stat + syst)
ATLAS-1 = 8 TeV, 20.3 fbs
WZ Selection
[TeV]jjm1.5 2 2.5 3 3.5
Sig
nific
ance
2−1−0123
35 / 35 Alex Martyniuk BSM searches at ATLAS
Did we find any errors?
In so many cross-checks, yes,bugs were found and fixed
Always attempted to shield thedata from any possible biasesby using control regions to test
After almost a year and a halfof scrutiny by theanalysers/Exoticsgroup/ATLAS we found nomajor issues
Therefore we continue topublish the final results
Run-2 was looming!
35 / 35 Alex Martyniuk BSM searches at ATLAS
X →WW limits
As no significant deviation was observed, we continue to set limits on theobserved distributions
95% confidence limits set on σ × B using the CLS prescription taking intoaccount the systematic uncertainties and background fit
[TeV]RS
G m1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
WW
) [fb
]→
RS
BR
(G×
) R
S G
→(p
p σ
1−10
1
10
210
310
410 Observed 95% CL
Expected 95% CL
uncertaintyσ 1±
uncertaintyσ 2±
= 1PIM k/RSBulk G
ATLAS-1 = 8 TeV, 20.3 fbs
Expected limits broadlyagree with the observedlimits
Exclusion of gravitonproduction at no masses
Deviation from thebackground observable ataround 2.1 TeV
Benchmark BulkRandall-Sundrum gravitonσ × B shown forcomparison purposes
35 / 35 Alex Martyniuk BSM searches at ATLAS
X → ZZ limits
As no significant deviation was observed, we continue to set limits on theobserved distributions
95% confidence limits set on σ × B using the CLS prescription taking intoaccount the systematic uncertainties and background fit
[TeV]RS
G m1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
ZZ
) [fb
]→
RS
BR
(G×
) R
S G
→(p
p σ
1−10
1
10
210
310
410 Observed 95% CL
Expected 95% CL
uncertaintyσ 1±
uncertaintyσ 2±
= 1PIM k/RSBulk G
ATLAS-1 = 8 TeV, 20.3 fbs
Expected limits broadlyagree with the observedlimits
Exclusion of gravitonproduction at no masses
Broad deviation from thebackground observable ataround 2 TeV
Benchmark BulkRandall-Sundrum gravitonσ × B shown forcomparison purposes
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? ATLAS searchesP
resc
ale-
wei
ghte
d ev
ents
1
10
210
310
410
510
610
710
810
910
[dat
a-fit
]/fit
-1
0
1
[TeV]jjReconstructed m0.3 0.4 0.5 1 2 3 4 5
Sig
nif.
-2
02
ATLAS-1L dt=20.3 fb∫=8 TeV, s
DataFit
*, m = 0.6 TeVq*, m = 2.0 TeVq*, m = 3.5 TeVq
ATLAS resolved dijetsearch [arXiv:1407.1376],nothing seen 4
ATLAS resolved Run-2dijet searchATLAS-CONF-2015-042,nothing seen 4
ATLAS semi-leptonicsearch W (lν)Z (jj)[arXiv:http://1503.04677],using similar BDRS-A CA1.2 reconstruction, intail/nothing seen 4
ATLAS semi-leptonicsearch W (jj)Z (ll)[arXiv:1409.6190], usingsimilar BDRS-A CA 1.2reconstruction, intail/nothing seen 4
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? ATLAS searches
[TeV]jjReconstructed m2 3 4 5 6 7
Even
ts
1
10
210
310
410
|y*| < 0.6Fit Range: 1.1 - 5.3 TeV
-value = 0.79p
[TeV]jjm2 3 4 5 6 7
Sign
if.
3−2−1−0123
ATLAS Preliminary-1=13 TeV, 80 pbs
DataBackground fitBumpHunter intervalBlackMax, m = 4.0 TeVBlackMax, m = 5.0 TeV
ATLAS resolved dijetsearch [arXiv:1407.1376],nothing seen 4
ATLAS resolved Run-2dijet searchATLAS-CONF-2015-042,nothing seen 4
ATLAS semi-leptonicsearch W (lν)Z (jj)[arXiv:http://1503.04677],using similar BDRS-A CA1.2 reconstruction, intail/nothing seen 4
ATLAS semi-leptonicsearch W (jj)Z (ll)[arXiv:1409.6190], usingsimilar BDRS-A CA 1.2reconstruction, intail/nothing seen 435 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? ATLAS searchesE
vent
s / 1
00 G
eV
-110
1
10
210
310
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxx
xxx
xxx
xxx
xxxxxxxxxx
xx
xx
xx
xxxxxxxxxxxxx
xxx
xxx
xxxxxxxxxxxxxxxx
xxx
xxx
xxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
ATLAS-1 Ldt = 20.3 fb∫ = 8 TeV, s
1 large-R jet≥ + ν l→W
DataW/Z+jets
+single topttMultijet
DibosonUncertainty
xxxx
xxxx
xxxx
xxxx
xxxx
G*(1200 GeV)
W'(1200 GeV)
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxxx
xxxx
xxxx
xxxx
xxxx
xxxxx
xxxxx
xxxxx
xxxxx
xxxxx
xxxxx
xxxxx
xxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxx
xxxxxxxxx
xxxxxxxxx
xxxxxxxxx
xxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
xxxxxxxxxxx
[GeV]lvJm800 1000 1200 1400 1600 1800 2000 2200 2400Dat
a / B
kg
00.5
11.5
2
ATLAS resolved dijetsearch [arXiv:1407.1376],nothing seen 4
ATLAS resolved Run-2dijet searchATLAS-CONF-2015-042,nothing seen 4
ATLAS semi-leptonicsearch W (lν)Z (jj)[arXiv:http://1503.04677],using similar BDRS-A CA1.2 reconstruction, intail/nothing seen 4
ATLAS semi-leptonicsearch W (jj)Z (ll)[arXiv:1409.6190], usingsimilar BDRS-A CA 1.2reconstruction, intail/nothing seen 4
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? ATLAS searchesE
vent
s / G
eV
-410
-310
-210
-110
1
10
210
310 DataZ+jetsZZ/ZW/WW
+Single TopttSys+Stat UncertaintyG*, m=800 GeV
10.0× Nominalσ
ATLAS = 8 TeVs
-1L dt = 20.3 fb∫ Res. Region
THigh-p
Channelµµ ee, →Z
[GeV]lljjm0 500 1000 1500 2000 2500
Dat
a / B
G
0.60.8
11.21.4
ATLAS resolved dijetsearch [arXiv:1407.1376],nothing seen 4
ATLAS resolved Run-2dijet searchATLAS-CONF-2015-042,nothing seen 4
ATLAS semi-leptonicsearch W (lν)Z (jj)[arXiv:http://1503.04677],using similar BDRS-A CA1.2 reconstruction, intail/nothing seen 4
ATLAS semi-leptonicsearch W (jj)Z (ll)[arXiv:1409.6190], usingsimilar BDRS-A CA 1.2reconstruction, intail/nothing seen 4
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? Across the ring....CMSCMS dijet search[arXiv:1501.04198], blip in2-btag at 2 TeV? Trick ofthe eye? 4
CMS run-2 dijet search[CMS-PAS-EXO-15-001],nothing 4
CMS WR search[arXiv:1407.3683], excessat 2 TeV in eejj channelonly 42�
CMS WW/WZ/ZZsemi-leptonic search[arXiv:1405.3447], intails/nothing seen 4
CMS WW/WZ/ZZ fullyhadronic search[arXiv:1405.3447], usesn-subjettiness, broad blipat ≈ 1.8TeV 42�
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? Across the ring....CMS
Dijet Mass (GeV)1500 2000 2500 3000 3500 4000 4500 5000 5500
[pb
/ GeV
]jj
/ dm
σd
-510
-410
-310
-210
-110
1data
background fit to data
QCD MC
q* (4.5 TeV)
| < 1.3ηΔ| < 2.5, |η|> 1.1 TeVjjM
Wide Jets
(13 TeV)-142 pb
CMSPreliminary
Dijet Mass [GeV]1500 2000 2500 3000 3500 4000 4500 5000 5500
σ(D
ata-
Fit)/
-3-2-10123
CMS dijet search[arXiv:1501.04198], blip in2-btag at 2 TeV? Trick ofthe eye? 4
CMS run-2 dijet search[CMS-PAS-EXO-15-001],nothing 4
CMS WR search[arXiv:1407.3683], excessat 2 TeV in eejj channelonly 42�
CMS WW/WZ/ZZsemi-leptonic search[arXiv:1405.3447], intails/nothing seen 4
CMS WW/WZ/ZZ fullyhadronic search[arXiv:1405.3447], usesn-subjettiness, broad blipat ≈ 1.8TeV 42�
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? Across the ring....CMSCMS dijet search[arXiv:1501.04198], blip in2-btag at 2 TeV? Trick ofthe eye? 4
CMS run-2 dijet search[CMS-PAS-EXO-15-001],nothing 4
CMS WR search[arXiv:1407.3683], excessat 2 TeV in eejj channelonly 42�
CMS WW/WZ/ZZsemi-leptonic search[arXiv:1405.3447], intails/nothing seen 4
CMS WW/WZ/ZZ fullyhadronic search[arXiv:1405.3447], usesn-subjettiness, broad blipat ≈ 1.8TeV 42�
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? Across the ring....CMSCMS dijet search[arXiv:1501.04198], blip in2-btag at 2 TeV? Trick ofthe eye? 4
CMS run-2 dijet search[CMS-PAS-EXO-15-001],nothing 4
CMS WR search[arXiv:1407.3683], excessat 2 TeV in eejj channelonly 42�
CMS WW/WZ/ZZsemi-leptonic search[arXiv:1405.3447], intails/nothing seen 4
CMS WW/WZ/ZZ fullyhadronic search[arXiv:1405.3447], usesn-subjettiness, broad blipat ≈ 1.8TeV 42�
35 / 35 Alex Martyniuk BSM searches at ATLAS
Are we alone? Across the ring....CMSCMS dijet search[arXiv:1501.04198], blip in2-btag at 2 TeV? Trick ofthe eye? 4
CMS run-2 dijet search[CMS-PAS-EXO-15-001],nothing 4
CMS WR search[arXiv:1407.3683], excessat 2 TeV in eejj channelonly 42�
CMS WW/WZ/ZZsemi-leptonic search[arXiv:1405.3447], intails/nothing seen 4
CMS WW/WZ/ZZ fullyhadronic search[arXiv:1405.3447], usesn-subjettiness, broad blipat ≈ 1.8TeV 42�
35 / 35 Alex Martyniuk BSM searches at ATLAS