1. No category

LHC-ATLAS実験における
ZHチャンネルを用いた
ヒッグス粒子のインビジル崩壊の探索
日本物理学会、佐賀大学、2014年9月19日
大川英希, 他ATLAS Collaboration
筑波大学 数理物質系・数理物質融合科学センター
標準理論のヒッグス粒子か?
(stat.)
ATLAS Prelim.
inc.
(sys
theory )
mH = 125.5 GeV
(theory)
H
µ = 1.57+0.33
-0.28
H
ZZ*
H
WW*
+ 0.35
- 0.32
+ 0.20
- 0.13
+ 0.17
- 0.10
l l
µ = 1.00+0.32
+ 0.21
- 0.21
+ 0.24
- 0.19
+ 0.16
- 0.08
-0.29
Combined
H
, ZZ*, WW*
+0.21
µ = 1.35
-0.20
W,Z H
bb
µ = 0.2+0.7
Total uncertainty
± 1 on µ
+ 0.23
- 0.22
+ 0.24
- 0.18
+ 0.17
- 0.12
4l
µ = 1.44+0.40
-0.35
Events / 0.1
Phys. Lett. B 726 (2013), 120
2500
Data
+
JP = 0
Background
2000
1500
1000
ATLAS
+ 0.14
- 0.14
+ 0.16
- 0.14
+ 0.13
- 0.11
500
± 0.5
0
± 0.4
s = 8 TeV
0.1
0.2
(8 TeV data only)
+0.5
µ = 1.4
-0.4
Combined
H b b,
+0.36
µ = 1.09
-0.32
Combined
+0.18
µ = 1.30
-0.17
+ 0.3
- 0.3
+ 0.4
- 0.3
+ 0.2
- 0.1
+ 0.24
- 0.24
+ 0.27
- 0.21
+ 0.08
- 0.04
+ 0.12
- 0.12
+ 0.14
- 0.11
+ 0.10
- 0.08
s = 7 TeV Ldt = 4.6-4.8 fb-1
s = 8 TeV Ldt = 20.3 fb-1
-0.5 0
0.5
1
1.5
2
Signal strength (µ)
ATLAS-CONF-2014-009 (2014)
大川英希
-1
L dt = 20.7 fb
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
|cos *|
-0.6 <0.1
H
H
•
現時点でヒッグス粒子の観測・測定は、標準理論
と矛盾しない（シグナルの強度・スピン）。
•
LHCではヒッグスの崩壊幅を直接測定できない。
間接測定でも崩壊幅の上限は標準理論の4∼5倍。 → 標準理論を超えた物理が、ヒッグスセクターに
存在する可能性が、現時点で棄却できない。
日本物理学会、2014年9月19日
2
標準理論を超えた物理(BSM)の探索
ヒッグスセクターでのBSM探索の方法:
ATLAS Simulation Preliminary
s = 14 TeV: Ldt=300 fb-1 ; Ldt=3000 fb-1
•
•
ヒッグス粒子の結合定数の精密測定 → 多くのモデルで数%以下の精度が必要。HLLHCのデータ(∼3000 fb-1)やILCを要する 。
gZ
ATL-PHYS-PUB-2013-014
WZ
Energy Frontier
ヒッグス粒子のBSM崩壊の直接探索:
暗黒物質へのインビジブル（非可視）崩壊など.
tg
to be in this situation, in which the picture of the Higgs boson may be very di↵erent from that in the
Z
ut, since the other particles in the sector are heavy, it is difficult to conclude this except by precision
重いヒッグス粒子の探索
rement.
•
µZ
al sizes of Higgs boson coupling modifications are shown in Table 3-1. More details of these estimates
ven in [23].
BSMによる結合定数のずれ
gZ
Model
V
b

Singlet Mixing
2HDM
⇠ 6%
⇠ 1%
⇠ 6%
⇠ 10%
⇠ 6%
⇠ 1%
Decoupling MSSM
Composite
Top Partner
⇠
0.0013%
⇠ 3%
⇠
2%
⇠ 1.6%
⇠ (3 9)%
⇠
2%
Snowmass, Energy Frontier Report, 2013
< 1.5%
⇠ 9%
⇠ +1%
Z
0.78
(Z )Z
0
able 3-1. Generic size of Higgs coupling modifications from the Standard Model values in classes of new
ysics models: mixing of the Higgs boson with a singlet boson, the two-Higgs doublet model, the Minimal
日本物理学会、2014年9月19日
大川英希
persymmetric
Standard Model, models with a composite Higgs boson, and models with a heavy vectorlike
0.1
0.2
=
XY
0.3
X
Y
3
ヒッグスのインビジブル崩壊
ヒッグス粒子と暗黒物質との相互作用による崩壊。超対称性などからも予想。
ZH随伴過程
暗黒物質？
Vector-Boson Fusion
q
q
χ
W/Z
H
W/Z
q
χ
q
•
ZH随伴過程及びvector-boson fusionのチャンネルは、特にインビジブ
ル崩壊への感度が高い。
•
Z(→ll)Hチャンネルは、バックグラウンドの評価が比較的し易い、 クリーンなチャンネル。
大川英希
日本物理学会、2014年9月19日
4
解析の方針
“Z+Missing ET (ETmiss)”
を用いた探索
Events
Phys. Rev. Lett. 112, 201802 (2014)
108
ATLAS
107
s = 8 TeV,
→ ℓℓ
106
L dt = 20.3 fb -1
105
→ ℓ ℓℓ
104
→ ℓℓ
103
→ ℓℓ
ℓ
→
102
10
Missing ET: ビームラインに垂直な平面
における運動量の不均衡
•
Data / MC
1
1.5
1
0.5
0
50
100
150
200
250
300
350
400
450
500
クリーンなチャンネルだが、シグナルの感度を下げず (ETmissのカットを
上げ過ぎず)に、Zバックグラウンドをいかに抑えるかがポイント。
大川英希
日本物理学会、2014年9月19日
5
Events / /8
イベントセレクション
107
6
ATLAS
s = 8 TeV,
10
L dt = 20.3 fb
-1
•
シングルレプトン(e, μ)・
ダイレプトントリガーで事象を選択。
•
Zボソン由来のe, μを選択;
レプトンは３つ以上存在しない。
(pT>7 GeV)
•
•
dϕ(l,l) < 1.7
→ ℓℓ
105
→ ℓ ℓℓ
→ ℓℓ
ℓ
ETmiss>90 GeV
4
10
→ ℓℓ
3
→
10
102
10
Data / MC
1
2
ジェット(pT>25 GeV)が事象中に存在し
ない。
1
0
0
0.5
1
1.5
2
2.5
3
pTmiss: 内部飛跡検出器のトラック
から再構成したMissing ET
ϕ
ETmissを用いたZバックグラウンドの抑制
dϕ(ETmiss,pTmiss) < 0.2
ETmiss > 90 GeV
dϕ(Z, ETmiss) > 2.6
|ETmiss - pTll| / pTll < 0.2
•
•
大川英希
•
•
日本物理学会、2014年9月19日
6
Not reviewed, fo
368
369
370
371
search channel via the H ! ZZ decay, especially for Higgs boson m
At the LHC, the dominant ZZ production mechanism is from qu
lesser extent from gluon-gluon fusion. Figures 1 and 2 shows the l
ZZ production with qq¯ and gg initial states.
バックグラウンド
BGの大きさ
•
Z
ZZ(→l+l-vv): qシグナルとの区別が極め
q
Z
q¯
Z
て困難. モンテカルロ (MC)で評価。
•
WZ: MCで評価。3-レプトンコント
q¯
Z
(a) u channel ZZ production.
ロール領域(CR)で確認。
(b) t channel ZZ production.
•
W+W-/tt&Wt/Z(→τ+τ-): e-μ CRを用いて、データから評価。
•
•
Z+jets: ABCD method(後述)を用いて、データから評価。適化でここま
Figure 1: The SM tree-level Feynman diagrams for ZZ production th
イベントセレ
colliders.
クションの最
で抑えた
W+jets/multijet: H→WW解析などで用いられている手法によ
g
り、データを用いて評価。ほぼ無視できる。
大川英希
日本物理学会、2014年9月19日
Z
7
Z バックグラウンド
サイドバンド
領域D
Events / 20 GeV
サイド
バンド
領域C
1012
ATLAS
s=8 TeV, L dt = 20.3 fb-1
ZH → ℓℓ + inv.
1010
miss
∆ϕ(E T
Data
Z
ee, µ µ
Other BG
Top quark
Sideband Region
108
miss
, pTmiss) > 0. 2, |E T
− pTℓℓ|/p ℓℓ
T > 0. 2
WW
WZ → ℓνℓℓ (incl.τ)
106
ZZ → ℓℓνν (incl.τ)
4
ZH → ℓℓ + inv., BR(H → inv.) = 1
10
102
1
シグナル
領域A
サイドバンド
領域B
Data / MC
|ETmiss - pTll| / pTll
ABCDメソッド
4
2
0
dϕ(ETmiss,pTmiss)
相関の極めて小さい２つの変数に対して、シグ
ナル領域とサイドバンド領域を定義する手法。
NAest = NBobs
obs
NC
obs
ND
2011
2012
50
100
150
200
250
E miss
[GeV]
T
Z BG
0.13 ± 0.12 (stat) ± 0.07 (sys)
0.9 ± 0.3 (stat) ± 0.5 (sys)
系統誤差は、NA/NB & NC/NDのカット依存性 、
Z以外のMCの差し引きによる不定性など
NC/ND~0.1, α=1.07 (2011), 1.04 (2012)
大川英希
日本物理学会、2014年9月19日
8
+
+
W W /Top/Z(→τ τ ):
Events / 10 GeV
•
W+W-/ttのダイレプトン崩壊事象&Wt/Z(→τ+τ -)事象におけるレプトンの
フレーバー対称性を活用。
e-μ CRから、データを用いてBGを評価。
BG,est.
Nee
=
BG,est.
Nµµ
=
1
2
1
2
data,sub
Neµ
k
data,sub
Neµ
1
k
k=
data
Nee
data
Nµµ
ATLAS
4
10
Preliminary
s = 8 TeV
eµ events
103
-1
Data L dt=13.0 fb
Z
Multijet
W
Top
WW
WZ
ZZ
*
*
SM H
ZZ , H WW
102
10
Data / MC
•
eμ評価法
1.2
1
0.8
•
NeμBG, est: BGの評価。
•
Neμdata,sub: CRにおけるデータからからWW/Top/Z (→τ+τ-)以外のBGを差し引いた数。
•
k-efficiency factor: 電子とミューオンの性能の違いを補正する因子
0
50
100
150
200
250
300
Emiss
[GeV]
T
WW/Top/Z(→τ+τ-)
大川英希
2011
2012
0.5 ± 0.4 (stat) ± 0.1 (sys)
20 ± 3 (stat) ± 5 (sys)
日本物理学会、2014年9月19日
9
105
ATLAS
Data
-1
s = 8 TeV, L dt = 20.3 fb
ℓℓ + e/ µ
104
Z boson
WZ → ℓνℓℓ (incl.τ )
ZZ → ℓℓνν, 4ℓ (incl.τ )
102
10
ZZは、4-レプトンCRの統計が不足。ZZ
の測定された断面積測定とNLOの理論計
算は現時点で一致。
1.5
0.5
0
(ZZ) = 7.1+0.5
0.4 (stat.) ± 0.3(syst) ± 0.2(lumi.)pb
NLO
(ZZ) = 7.2+0.3
0.2 pb
measured
1
50
100
150
200
250
300
350
400
450
miss
ET
500
[GeV]
レプトンの不定性1.0-1.5%.ジェット由来の不定性
3-6%。
PDF, スケールの不定性を、ETmiss分布の形への影響も
含めて考慮。 規格化への影響は、5%程度。
gg→WW/ZZ→l+l-vvの寄与も考慮 (qq→ZZの~3%)。
大川英希
ATLAS-CONF-2013-020
1.3
s= 7 TeV
CT10 central value / Mid-point
MSTW central value / Mid-point
1.2
NNPDF central value / Mid-point
1.1
1
0.9
0.8
LHC Higgs Yellow Report II
Pdf+Alpha Uncertainty
日本物理学会、2014年9月19日
LHC HIGGS XS WG 2011
Data / MC
1
•
•
WW
10
•
WZ/ZZ共に、MCで評価。WZは、3-レプ
トンCRで確認。
Top quark
3
•
•
1 + δenv (NLO)
Events
ZZ/WZ バックグラウンド
s
0.7
100 150 200 250 300 350 400 450 500 550
10600
M4l [GeV]
Events / 30 GeV
インビジブル崩壊探索の結果
3
10
102
標準理論の予想からの有意なずれは
ATLAS
Data
s = 8 TeV, L dt = 20.3 fb-1
ZH → ℓℓ + inv.
ZZ → ℓℓνν (incl. τ)
観測されなかった
WZ → ℓνℓℓ (incl. τ)
WW, dilep. t¯
t, Wt, Z → ττ
Z
ee, µ µ
W + jets, multijet, semilep. top
10
ZH → ℓℓ + inv., BR(H → inv.) = 1
•
インビジブル崩壊の分岐比に、75%
1
Data / Expected
ETmissを識別変数として、ヒッグスの
observed (63% expected) @ 95% CL
のちのCMSの結果よりも
の制限を与えた。
2
1.5
1
0.5
20%良い制限
100
150
200
250
300
350
400
450
miss
E T [GeV]
•
LHCで初めて得られた、ヒッグスの
インビジブル崩壊の分岐比の制限。
4
Data Period
2011 (7 TeV)
2012 (8 TeV)
ZZ → !!νν
20.0 ± 0.7 ± 1.6
91 ± 1 ± 7
W Z → !ν!!
4.8 ± 0.3 ± 0.5
26 ± 1 ± 3
Dileptonic tt¯, W t, W W , Z → τ τ
0.5 ± 0.4 ± 0.1
20 ± 3 ± 5
Z → ee, Z → µµ
0.13 ± 0.12 ± 0.07
0.9 ± 0.3 ± 0.5
W + jets, multijet, semileptonic top
0.020 ± 0.005 ± 0.008 0.29 ± 0.02 ± 0.06
Total background
25.4 ± 0.8 ± 1.7
138 ± 4 ± 9
Signal (mH = 125.5 GeV, σSM (ZH), BR(H → inv.) = 100%)
8.9 ± 0.1 ± 0.5
44 ± 1 ± 3
Observed
28
152
TABLE I. Number of events observed in data 日本物理学会、2014年9月19日
and expected from the signal and from each background source for the 7 and 811
大川英希
TeV data-taking periods. Uncertainties on the signal and background expectations are presented with statistical uncertainties
Higgs-portalモデルによる解釈
Higgs-portalモデル：暗黒物質 (DM)がヒッグス粒子のみと相互作用。暗黒物質の反
応断面積が極めて小さくなり、既存の探索実験で未観測であることと整合。
100
LHCでの探索
χ
χ
暗黒物質の直接探索実験
χ
λhχχ
ヒッグス・暗黒物質
の結合
λhχχ
h
h
fN
χ
N
(a)
N
(b)
DM-核子の反応断面積
ヒッグスのインビジブル崩壊幅 ヒッグス-DMの結合定数
Figure 65: Feynman diagrams for the decay of the Higgs boson into dark matter particles (a) and scat2 exchange of a Higgs boson (b). The Higgs-dark
tering of dark matter particles off of a nucleon with the
matter interaction vertex has a coupling constant of λh
hχ χ . In the scattering diagram the Higgs-nucleon
(h
)
coupling BR(h
strength is
) =parameterized with a form factor, f N .
(h
)
(h
大川英希
) + (h
N
SM )
日本物理学会、2014年9月19日
12
DM−Nucleon cross seciton [cm2]
Higgs-Portalモデルによる解釈
10-37
10-38
10-39
10-40
10-41
10-42
10-43
10-44
10-45
10-46
10-47
10-48
10-49
10-50
10-51
Higgs-portal Model
s = 7 TeV, ∫ Ldt=4.5 fb-1
s = 8 TeV, ∫ Ldt=20.3 fb -1
ZH → ℓℓ + inv.
スカラー
マヨラナ
ベクター
DAMA/LIBRA 3σ
CDMS 95% CL
XENON10
LUX
ATLAS, vector DM
1
•
ATLAS
10
CRESST 2σ
CoGeNT
XENON100
ATLAS, scalar DM
ATLAS, fermion DM
102
ATLASの結果のバンド
は、ヒッグス-核子間の
form factorの不定性に
よるもの
103
DM Mass [GeV]
ヒッグスのインビジブル崩壊比への制限は、ヒッグスと暗黒物質との結
合定数への制限を与える。Higgs-portalモデルでは、MDM<MH/2におい
て、LHCが突出した感度を持つ。
大川英希
日本物理学会、2014年9月19日
13
今後の展望
•
14 TeV LHC及びHigh Luminosity LHC (HL-
arXiv:1309.7925, ATL-PHYS-PUB-2013-014
H. Okawa, Moriond QCD 2014での発表
LHC)でのインビジブル崩壊への感度について
は、SnowmassやECFA (2013)に向けて、既
に研究を行った。
•
ZHチャンネル単独であっても、 3000 fb-1では
BR Limit
ATLAS
CMS
300 fb-1
[23,32]%
[17,28]%
3000 fb-1
[8,16]%
[6,17]%
BR(H→inv)<~10%に感度がある。超対称性や
他のBSMモデルにとっても非常に重要。
•
VBFや他の探索チャンネルや又カプリング測定と合わせると更に感度が上がる。
•
今後は、他のチャンネルでのコンビネーションや、Run 2でのパイルアップの対策や
ZH探索でのバックグラウンドの抑制・評価の改善等に取り組む。
大川英希
日本物理学会、2014年9月19日
14
まとめ
•
ヒッグス粒子が標準理論の枠内にあるかどうかを探るために有用な方法
の一つである、インビジブル崩壊の探索をZHチャンネルを用いて行っ
た。
•
標準理論からの有意なずれは観測されなかった。LHCで初めて、ヒッグ
ス粒子のインビジブル崩壊比に対して制限を与えた。
BR(H→inv.) < 75% (obs), 63% (exp) @ 95%CL。
•
インビジブル崩壊比に対する制限を、Higgs-portal暗黒物質モデルにお
いて解釈した。暗黒物質の質量がヒッグス粒子の半分以下である場合に
は、突出した感度を持ち、直接探索実験と相補的な結果が得られた。
•
Run-2に向けて、更なる改善・探索を行う。
大川英希
日本物理学会、2014年9月19日
15
バックアップ
大川英希
日本物理学会、2014年9月19日
16
EW for qq¯ → ZZ
7.1
Parton showering
7.1
Z BG systematic
7.4
Luminosity
7.3
Electron energy scale
7.1
Electron ID efficiency
7.1
Muon reconstruction efficiency
7.1
Jet energy scale
7.1
Sum of remaining systematic uncertainties
7.1
s = 8 TeV
25
ATLAS Simulation Preliminary
All systematic
9.9
ATLAS Preliminary
gg
ZZ
2e2µ
H
ZZ
2µ2
No
systematic
7.1
gg H*
ZZ (S)
10-1
10-2
gg
Events / 30 GeV
d /dm4l [fb/GeV]
ヒッグスの崩壊幅の間接測定
ZZ (B)
20
ATLAS-CONF-2014-042
Data
gg+VBF
qq
(H* ) ZZ
ZZ
WZ
Z( ee/ µ µ )+jets
s = 8 TeV: Ldt = 20.3 fb-1
(H* upper
) ZZ
Table 6: The expected 95%ggCL
limit on µoff-shell in the 2!2ν channel, with
a ranked listing of
Other backgrounds
-3
gg (H* ) ZZ (µ
=10)
All contributions (µ
=10)
each 10
systematic uncertainty individually, and comparing15to including no systematic
uncertainty
or all
Stat.+syst. uncertainties
B
systematic uncertainties. The upper limits are evaluated using the CL s method, assuming RH
∗ =1.
off-shell
off-shell
10-4
10
10-5
5
10-6
200
B
RH
∗
400
600
400
800
1000
m4l [GeV]
Observed
0.5
1.0
2.0
WW/Top/Z
500
600
700
mT [GeV]
Median expected
0.5
1.0
2.0
µoff-shell
5.6
6.7
9.0
6.6
7.9
10.7
ΓH /ΓSM
H
4.1
4.8
6.0
5.0
5.8
7.2
ΓH /ΓSM
H
4.8
5.7
7.7
7.0
8.5
12.0
Alternative hypothesis
B
RH
∗ = 1, µoff-shell = 1
B
SM = 1, µ
RH
∗ = 1, Γ H /Γ H
on-shell = 1.51
B
SM = 1, µ
RH
∗ = 1, Γ H /Γ H
on-shell = 1
SM within the range of
Table
7:
The
observed
and
expected
95%
CL
upper
limit
on
µ
and
Γ
/Γ
off-shell
H
日本物理学会、2014年9月19日
H
大川英希
B
0.5 < R ∗ < 2, combining the ZZ → 4! and ZZ → 2!2ν channels. The bold numbers correspond to the
17
ヒッグスの崩壊幅と生成断面積
i
•
•
•
•
•
大川英希
H
f
2
i
2
f
H
σi→H→f: ヒッグス生成過程i、崩壊過程fの反応断面積
κi: ヒッグスのiへの結合定数
κi: ヒッグスのfへの結合定数
ΓH: ヒッグスの崩壊幅
ΓHとκi2κf2が同じ比率で標準理論の予測に対してスケールすれば、
ヒッグスのシグナルの強度は、標準理論と変わらない。
日本物理学会、2014年9月19日
18
tegrated dataset of 300 fb 1 (left) and 3000 fb
wo uncertainty scenarios described in the text.
1
(right).
Projections
CMS NOTE-13-002
CMS Projection
rio 1
eory Unc.
Expected uncertainties on
Higgs boson couplings
3000 fb-1 at
s = 14 TeV Scenario 1
3000 fb-1 at
s = 14 TeV No Theory Unc.
κγ
Scenario 1:
same systematics
as Run 1
κW
κZ
Same systematics
as Run 1, but w/o
theory unc.
κg
κb
κt
κτ
5
ertainty
0.00
0.05
0.10
0.15
expected uncertainty
日本物理学会、2014年9月19日
signal
coupling modifiers (right).
大川英希 strengths (left) and
19
Scenario I
µˆ w/Theory error
µˆ wo/Theory error
Significance
µˆ w/Theory error
µˆ wo/Theory error
+0.56
+0.54
3.2
+0.33
+0.32
0.54
0.54
0.19
0.18
+0.20
+0.18
-
+0.19
+0.18
6.4
+0.18
+0.16
0.19
0.17
Projections: H→bb
Scenario II
0.32
0.32
0.17
0.16
Table 12: Expected signal sensitivity as well as the precision on the signal strength measurement for
mH = 125 GeVfor the one-lepton, two-lepton and combined searches with 3000 fb 1 with hµipuATL-PHYS-PUB-2014-011
= 140.
Stat-only
Theory-only
Scenario I
Scenario II
大川英希
Significance
µˆ Stats error
µˆ Theory error
Significance
µˆ w/Theory error
µˆ wo/Theory error
Significance
µˆ w/Theory error
µˆ wo/Theory error
One-lepton
5.5
+0.18 0.18
+0.08 0.05
1.8
+0.57 0.57
+0.56 0.57
2.1
+0.48 0.47
+0.46 0.46
Two-lepton
4.6
+0.23 0.22
+0.08 0.06
3.5
+0.30 0.29
+0.29 0.29
-
One+Two-lepton
7.1
+0.14 0.14
+0.09 0.06
3.9
+0.27 0.26
+0.26 0.26
4.1
+0.26 0.25
+0.25 0.24
Table 13: Expected signal sensitivity as well as the precision on the signal strength measurement for
mH = 125 GeVfor the one-lepton, two-lepton and combined searches with 300 fb 1 and hµipu = 60 after
One-lepton
Two-lepton One+Two-lepton
including the perspective of a more performant
analysis.
Stat-only
Significance
15.4
11.3
19.1
µˆ Stats
error of+0.07
+0.09
0.09 with+0.05
0.05significance
1 it can
to be +0.27
. With an integrated
luminosity
3000 fb 0.06
be observed
an expected
0.26
+0.09 0.07 +0.07 0.08
+0.07 007
of 8.8 andTheory-only
µˆ = ±0.14. µˆ Theory error
Significance
2.7
8.4
8.8
Scenario I
µˆ w/Theory error +0.37 0.36 +0.15 0.15
+0.14 0.14
+0.12 0.12
11 Conclusion µˆ wo/Theory error +0.36 0.36 +0.14 0.12
Significance
4.7
9.6
II production
µˆ w/Theory error
+0.23 with
0.22 leptonically
- decaying+0.13
A study of Scenario
Higgs boson
in association
W and 0.13
Z bosons using
µ
ˆ
error
+0.21
0.21
+0.11
0.11
wo/Theory
parameterised functions to model
the behaviour of the upgraded ATLAS detector at the high-luminosity
p
LHC with s = 14 TeV has been performed.
Table 14: Expected signal sensitivity as well as the precision on the signal strength measurement for
1
mH = 125the
GeVfor
the one-lepton,
two-lepton
and[11],
combined
searches
3000 fb
with
140
Following
analysis
strategy described
in Ref.
and using
onlywith
the decay
modes
ofhµi
thepuW= and
日本物理学会、2014年9月19日
including
thetoperspective
of a electron
more performant
Zafter
bosons
leading
a high energy
or muon analysis.
in the final state, we obtain expected sensitivities
20
jet
! miss
! miss
Fig. 4. The azimuthal separation between the missing transverse momentum vector, p
, and the nearest jet in the event !φ( p
, p! T ) after the E Tmiss requirement, for
T
T
the high mH search region. Figure (a) refers to the low pile-up data and figure (b) to the high pile-up data.
H→ZZ→llvv探索 (2011)
Table 1
The expected number of background and signal events along with the observed numbers of candidates in the data, separated into the low and high mH search regions and
Phys. Lett. B, 717 (2012) 29
the low and high pile-up periods. The quoted uncertainties are statistical and systematic, respectively.
Source
Low mH search
High mH search
Low pile-up data
High pile-up data
Low pile-up data
High pile-up data
Z
W
Top
Multijet
ZZ
WZ
WW
Total
Data
40.1 ± 5.0 ± 7.9
4.6 ± 2.2 ± 4.6
23.2 ± 1.3 ± 5.4
1.1 ± 0.2 ± 0.5
33.4 ± 0.7 ± 3.9
23.3 ± 1.0 ± 2.8
25.5 ± 0.8 ± 3.0
151 ± 6 ± 11
158
265 ± 13 ± 67
5.8 ± 1.8 ± 5.8
27.9 ± 1.3 ± 5.3
1.1 ± 0.2 ± 0.6
36.7 ± 0.7 ± 4.3
25.2 ± 1.0 ± 3.0
32.4 ± 0.9 ± 3.8
394 ± 13 ± 67
442
0.8 ± 0.3 ± 0.8
1.5 ± 0.8 ± 1.5
16.0 ± 1.1 ± 4.0
0 .1 ± 0.1 ± 0.0
28.4 ± 0.6 ± 3.4
17.1 ± 0.8 ± 2.1
9.4 ± 0.5 ± 1.1
73.3 ± 1.8 ± 6.1
77
11.6 ± 2.1 ± 2.9
2.2 ± 1.3 ± 2.2
17.2 ± 1.0 ± 3.9
0 .1 ± 0.1 ± 0.0
31.9 ± 0.7 ± 3.8
18.9 ± 0.8 ± 2.3
13.3 ± 0.5 ± 1.6
95.2 ± 2.9 ± 6.9
109
mH [GeV]
Signal expectation
16.4 ± 0.3 ± 2.9
14.4 ± 0.2 ± 2.5
6.2 ± 0.1 ± 1.1
2.7 ± 0.0 ± 0.5
17.5 ± 0.3 ± 3.1
15.4 ± 0.2 ± 2.7
6.5 ± 0.1 ± 1.1
2.9 ± 0.0 ± 0.5
Events / 30 GeV
ATLAS 2011
s = 7 TeV
11.1 ± 0.2 ± 1.9
Events / 50 GeV
10.3 ± 0.2 ± 1.8
200
300
400
500
600
ATLAS 2011
s = 7 TeV
400 samples: the first uses the dilepton mass side70
! miss
! jet
shown
independent control
,
p
), where
the inclusive Z boH ZZ ll
H inZZFig.ll4. At low !φ( p
-1
-1
T
data L dt = 2.4 fb
data L dtT= 2.4 fb
pile-up
pile-up data
350 atHigh
band and requires
least
onedata
identified b-jet
(Fig. 3(b)), while
son High
background
dominates, Total
a small
discrepancy is observed be60
Total BG
BG
Top
Top
the second selects
electron–
tween the data and the expected
backgrounds. This discrepancy is
300 events containing oppositely-charged
50
ZZ,WZ,WW
ZZ,WZ,WW
muon pairs (Fig. 3(c)).
within the systematic uncertainties applied on the inclusive Z boZ,W
Z,W
250
Additional backgrounds can arise from multijet
son
background.
Signal (m =events
200 GeV)or inSignal (m = 400 GeV)
40
H
H
200production due to heavy flavour decays or jets
clusive W boson
The signal efficiencies and overall background expectations are
30
misidentified as
leptons. The normalisation of the inclusive W bosimilar in the electron and muon channels; therefore only com150
son background
in this search is obtained from a control sample
bined results are presented. The numbers of candidate H → ZZ →
20
100
of events with a like-sign electron–electron or electron–muon pair
#+ #− ν ν¯ events selected in data and the expected yields from sig10
50of the dilepton mass distribution and with no
in the sidebands
nal and background processes are shown in Table 1.
miss
0 3(d) shows the E T
0
b-tagged jets. Fig.150
distribution
following
200
250
300
350
400
450this
200
300
400
500
600
700
日本物理学会、2014年9月19日
21
大川英希At large E Tmiss this distribution is dominated
by the inprocedure.
6. Systematic uncertainties
mT [GeV]
mT [GeV]
clusive W boson background.
Not reviewed, for internal circulation only
1238
Fjet
above
is ×
thef1fake-factor.
In contrast
(16)to the fake-rate in matrix method, now t
Nmulti−
= ∑
NFF
× f2
i listed
selection iof the fake-factor is orthogonal to the numerator selection. From now on, we de
se”. The vector on the
right
hand
side
of
the equation describes the inaccessible truth quantities
Nevents
object passing the tight selection as “Good” (mis-identified) lepton, while the lepton
i
i
i
selection
“Bad”
(anti-identified)
lepton,
and we
replace the symbols
with. “G”
NW +these
= ∑quantities
NRF
× r1i ×this
f2i + N
f1i N
×RR
r2i as
+does
NFF
×
f1i ×
f2i
Since
arejettruth
means
not
contain
the
other
components
like“T”NRF
jets&multi−
FR ×
i
with “B”. The fake-factor then is:
of this the same holds for the vector on the
left side of the(フェイクレプトンBG)
equation, which means e.g. NT T does
f ake
NGood
components
like NT L . This leads to the fact that the definition of “L”Fi ==
is “pass
loose selection
einfake-factor
method
f ake
N
Bad
This
is in contrast
the
selectionselection
Nloose to
instandard
which also all objects of the
dight
is in selection”.
fact a “mapping”
for lepton-like
object to
from
jetsbasic
from anti-lepton
The expression 24 now can be written as:
ection
are
contained.
n selection.
We
can
re-write
equation
14
as
below
(simply
take r1 = r2 = 1),
Fake-factor法
coefficients
f
fake-rate) denote the probability
in the loose selection is

that a true fake


i (called
N
1
f2
f1
f1 f2
NRR
NT T
i
i
i
i
i
i
i
N
=
N
×
F
+
N
×
F
−
2N
×
F
×
F
W
+
jets
∑
ucted as 
aN
signal-like
lepton
(“Tight“)
while
the
coefficients
r
represent
the
probability
that
a
real
GB
2
BG
1
BB
1
2

 

i
0
f1 (1 − f2 ) 
i
  NRF 
 T L  = 0 1 − f 2
(17)





 Nselection
N
the loose
is
reconstructed
as
a
tight
lepton.
The
indices
i
(i
=i 1, 2)
represent
leading
0
0
1
−
f
(1
−
f
)
f
N
1
1 2
FR
LT
i
i
N
=
N ×F ×F
0 lepton
0 (i =0 2) according
(1 − f1 )(1 − to
f2 )the N
NLL
FF
= 1) or subleading
lepton
pT .multi jet ∑i BB 1 2
1239
1240
W+jets/Multijet
1241
1242
1243
events
events
}
}
Nevents
ove equation we can see that the matrix elements which used to correct real lepton contribui
i
i
NW + jets&multi− jet = ∑ NGB
× F2i + NBG
× F1i − NBB
× F1i × F2i
f
ake
real
w set to be zero. Therefore these real lepton
correction terms have toNbe
replaced withi MC
N
データから
tight
tight
when dealing with control samples.
These corrections will評価する数
be applied
for the entries in the
f
=
r = real
(15)
The fake-factor method
only needs to deal with the fake-factor determination withou
f ake 1244
(Nij: lepton i,jのある事
N
the matrix which are related to real lepton
contribution
processes.
matrixreliably with small statistic control samp
Nloose
(R:dilepton
realレプトン,
loose ItThis
1245
leptonfrom
efficiency
determination.
also works
象数; L: looseレプト
n beン,
easily
solved and the expression for the background
estimation
then can
be towritten
in
1246
the matrix
method
normally
needs
have enough
statistics to get a robust estimation
F:
fakeレプトン)
T:
tightレプトン)
nverting
the
matrix
in
equation
14,
one
can
get
the
truth
variables
and
the
from
measurable quantities:
1247
select the fake-factor method as the primary method to contribution
estimate the W + jet
and multi-
publication. 16 where i runs over all selected di-lepton
nd Multijets can be estimated by usingforEquations
hich pass “Loose”Fake-factorのCRによる不定性、pileupによる不定性などを考慮。
lepton cuts and all other
ZH selection cuts.
12.6.4 Fake+factor measurement using Z+jet control sample
1248
•
1249
The fake-factor is measured via a Tag and Probe method using jet-enriched sample. We fi
1251
lepton-object
criteria, then describe the control sample selection.
W+jets/Multijet
Nevents selection
1252
The “Good”ilepton selection
sample
is the same as the full sele
i
ifrom thei jet-enriched
i
i
N2011
×Ther1(stat)
× ±f2lepton
+ NFR
× f × r2 jet-enriched sample is in fac
W+
1253jets
in =
the
ZH analysis.
“Bad”
selection
0.020
±NRF
0.005
0.008
(sys) 1from the
1254
selection i(or invert some lepton selection cuts) such that the prompt leptons from W an
2012
0.29 The
± 0.02
(stat) ±
0.06
1255
suppressed.
definitions
ofevents
“Good”(sys)
and “Bad” leptons are summarized in Table 3
N
i fake background
i
i estimation, since the faked o
1256
overlap removal with a jet is important for
(16)
N
=
N
×
f
×
f
multi−
jet overlap with jets.
FFIn the 1nominal
2 analysis, the selected
1257
less
isolated
and
usually
electron
日本物理学会、2014年9月19日 i
22
1258
overlaps with a selected jet with 0.2 < ∆R(e, jet) < 0.4, and the selected muon is remove
1250
∑
大川英希
∑
103
2
10
Data
ATLAS
ZZ → ℓℓνν (incl. τ)
s = 7 TeV, L dt = 4.5 fb-1
ZH → ℓℓ + inv.
WZ → ℓνℓℓ (incl. τ)
WW, dilep. t¯
t, Wt, Z → ττ
Z
10
ee, µ µ
W + jets, multijet, semilep. top
ZH → ℓℓ + inv., BR(H → inv.) = 1
Events / 30 GeV
Events / 30 GeV
ZHインビジブル結果
2
10
ATLAS
Data
s = 8 TeV, L dt = 20.3 fb-1
ZH → ℓℓ + inv.
ZZ → ℓℓνν (incl. τ)
WZ → ℓνℓℓ (incl. τ)
WW, dilep. t¯
t, Wt, Z → ττ
Z
10-1
ee, µ µ
W + jets, multijet, semilep. top
10
1
ZH → ℓℓ + inv., BR(H → inv.) = 1
2
1
100
150
200
250
300
350
400
450
miss
E T [GeV]
Data / Expected
1
10-2
3
Data / Expected
103
2
1.5
1
0.5
100
150
200
250
300
350
400
450
4
miss
E T [GeV]
Data Period
2011 (7 TeV)
2012 (8 TeV)
ZZ → !!νν
20.0 ± 0.7 ± 1.6
91 ± 1 ± 7
W Z → !ν!!
4.8 ± 0.3 ± 0.5
26 ± 1 ± 3
Dileptonic tt¯, W t, W W , Z → τ τ
0.5 ± 0.4 ± 0.1
20 ± 3 ± 5
Z → ee, Z → µµ
0.13 ± 0.12 ± 0.07
0.9 ± 0.3 ± 0.5
W + jets, multijet, semileptonic top
0.020 ± 0.005 ± 0.008 0.29 ± 0.02 ± 0.06
Total background
25.4 ± 0.8 ± 1.7
138 ± 4 ± 9
Signal (mH = 125.5 GeV, σSM (ZH), BR(H → inv.) = 100%)
8.9 ± 0.1 ± 0.5
44 ± 1 ± 3
Observed
28
152
TABLE
I. Number of events observed in data and
expected from the signal and from each background source for the 7 and
日本物理学会、2014年9月19日
238
大川英希
TeV data-taking periods. Uncertainties on the signal and background expectations are presented with statistical uncertainties
ZZの理論からの不定性
1.2
NNPDF central value / Mid-point
1.1
1.4
ss==77TeV
TeV
CT10 central value/ Mid-point
MSTW central value/ Mid-point
1.2
1.3
1.2
1.1
1.1
NNPDF
value/
Mid-point
QCD Scale
Q = central
mZ/2 (NLO
CT10)
QCD Scale Q = 2mZ (NLO CT10)
gg Ratio σ(Q) / σ(mZ)
MSTW central value / Mid-point
1.5
1.3
LHC HIGGS
HIGGS XS
XS WG
WG 2011
2011
LHC
CT10 central value / Mid-point
env
s= 7 TeV
NLO Ratio σ1(Q)
+ δ/ σ(mZ)
(gg)
1.3
LHC HIGGS XS WG 2011
1 + δenv (NLO)
LHC Higgs Yellow Report II
1
1
1
1
1
1
11
0
0.9
0.9
0.9
0.8
0.8
0.7
0.8
0
0.6
0
Pdf+Alpha Uncertainty
s
0.7
100 150 200 250 300 350 400 450 500 550 600
M4l [GeV]
0
Pdf+Alpha
Uncertainty
QCD
Scale Uncertainty
s
0.7
0.5
300350 400
400450 500
500 550 600
100100
150 200200
250 300
600
M
M4l4l[GeV]
[GeV]
0
QCD scale uncertainty is reduced for our llvv analysis due to:
1. Non-presence of Zγ contributions for ZZ→llvv
2. Our
ZZ section
samples
use
dynamic
instead
of the with
fixed
scale)
Fig. 98: The difference between the central value of the
cross
and
section
computed
3
Fig.
100:
The ratio
ofthe
thecross
crossscale
sections
computed
at different
(∗)
different PDF sets and the total PDF+αs variation (bluevalue
markers)
themfor
by qq
plus→and
qq →
ZZminus
→ 1σ
2e2for
µ (left)
and
gg →
of thevarying
QCD scale
24
大川英希
筑波大学、2014年4月8日
ZZ(∗) → 2e2µ (left) and for gg → ZZ(∗) → 2e2µ (right)7as
a function
of m2e2The
7 TeV
TeV
from MCFM.
linefrom
is theMCFM.
parametrisation describe
µ atred
ZH→ll+invisibleシグナル
•
HAWKというプログラムを用いて評価。
Higgs pT Uncertainty
•
ETmissの分布の不定性は、主にNLOのelectroweak correctionから。
0.1
ZH 8 TeV 2012
ZH 7 TeV 2011
0.08
0.06
0.04
0.02
100
150
200
250
300
Higgs pT
大川英希
筑波大学、2014年4月8日
350
400
450
Emiss
[GeV]
T
25
600
ATLAS
500
ZH ,SM
s = 7 TeV, L dt = 4.5 fb-1
× BR(H
s = 8 TeV, L dt = 20.3 fb-1
400
ZH
inv.) [fb]
リミット
300
ZH → ℓℓ + inv.
Observed 95% CL limit
Expected 95% CL limit
±1
±2
200
100
0
大川英希
150
200
250
300
日本物理学会、2014年9月19日
350
400
mH [GeV]
26
have reported an observation of a dark matter signal, including CRESST [62], DAMA [63], and
χ
NT [64]. The most recent observation from the CDMS collaboration [65] provides compellingχ
nce an 8.6 GeV dark matter particle. Not all of the observations are consistent with each other and
λhχχ
results are disputed by the community. Direct detection experiments make no a priori assumption
the mechanism by which dark matter particles interact with Standard Model particles, but it is
le that the interaction is through the exchange of a Higgs boson. If dark matter couples toλhχχ
the Stanh
h the Higgs mass then
Model through the Higgs boson and the mass of the particle is less than half
s to the dark matter particle will enhance the invisible branching fraction. Under the assumption
fN
ark matter couples to the Standard Model only through the Higgs boson we aim to place limits
ヒッグス-DMの結合定数
DM-核子の反応断面積
ヒッグスのインビジブル崩壊幅
imentary to the
direct detection results on the mass and interaction
cross section of the dark matter
χ
N
e.
(a)
(b)
ggs Portal models [66, 67, 68] make a simple, ad-hoc extension to the Standard Model by introg a new particle that couples to only the Higgs boson. The interaction strength is introduced with
(h
) Figure 65: Feynman diagrams for the decay of the Higgs boson into dark matter
)
=
pling constant, λhBR(h
.
Within
this
model
the
scattering
and decay process can be compared by exχχ
(h
) + (h tering
SM ) of dark matter particles off of a nucleon with the exchange of a Higgs boso
ng the limits in terms of this coupling constant. Figure 65 shows feynman diagrams for both the
matter
interaction
vertex
has arules
coupling
constant of λhχ χ . In the scattering diagra
and scattering processes where λhχ χ appears in both
diagrams.
Using the
feynman
for these
coupling
strength in
is terms
parameterized
with
a form factor, fN .
ms the Higgs partial width and scattering cross section
are determined
of λhχ χ . The
Higgs
width for the decay to dark matter particles for the scalar, vector, and fermion cases is given in
ions 22, 23, and 24 respectively.
インビジブル崩壊と暗黒物質
(h
Γ
Scalar
)
(h → χ χ ) =
λh2χScalar
v2
χ
64π mh
2
h
!
"
2mχ
1−
mh
#2 $1/2
!
"
#2 $1/2
2 Vector v2 %
&
λ
2mχ
hχ χ
4
2 2
4
1
−
m
−
4m
ΓVector (h → χ χ ) =
m
+
12m
χ h
χ
256π m4χ mh h
mh
ΓMajorana (h → χ χ ) =
2m
λh2χMajorana
v
h
χ
32π Λ2
!
"
2mχ
1−
mh
#2 $3/2
N
(22) =
σχScalar
N
σχVector
N(23) =
λh2χScalar
χ
m4N fN2
!
"
16π m4h mχ + mN 2
λh2χVector
χ
m4N fN2
"
!
16π m4h mχ + mN 2
2 Majorana
λ
m2χ m4N fN2
hχ χ
Majorana
σχ N (24) =
!
"
4π Λ2 m4h mχ + mN 2
he partial width is a function of only the Higgs
bosonThe
mass,
the section
dark matter
the vacuum
1474
cross
has mass,
an additional
dependence on the nucleon mass, mN an
tation value, and the coupling constant. Note the introduction of a cutoff scale, Λ in the fermionic
1475
which quantifies the coupling strength between the Higgs boson and the Nucleon.
In this case the Higgs interaction operator has dimension five and is non-renormalizable. A cutoff
1476
termined using lattice calculations and suffers from large theoretical uncertainties [
s added that assumes the presence of new physics at a higher energy scale which would produce a
ouplings to photons,  , and gluons, g , are introduced to absorb the possible
les through loops. The Higgs boson production modes are assumed to be the
ATLASの結合定数測定から
CONCLUSIONS
h2
P
i
=
i2
h / h,SM
=
=
0.0023 2
P
i
+
i2 /(1
BRi )
h / h,SM ,
is parametrized by
ATLAS-CONF-2014-010
i
width of the Higgs boson to the SM expectation,
-2 ln Λ(BR )
9
14
(18)
∫
s= 8 TeV, ∫ Ldt = 20.3 fb
10
= 125.5 GeV to photons and gluons are 0.0023
0.085 g2
+ 0.91,
12
h → γ γ , ZZ*, WW*, ττ, bb,
ATLAS Preliminary
s= 7 TeV,
Ldt = 4.6-4.8 fb
-1
miss
Zh → ll + ET
obs.
-1
:
exp.
h → γ γ , ZZ*, WW*, ττ, bb :
s of a Higgs boson with mh
[κ γ , κ g, BRi ]
obs.
exp.
8
d 0.91 is the sum of the branching ratios of the Higgs boson to massive partition and decay rates of all channels are fit with functions
of g ,  , and BRi .
6
plings are treated as nuisance parameters.
4
od scan as a function of BRi is shown in Fig. 6. There
is a lower physical
0, with the SM corresponding to BRi = 0. Ignoring
this boundary, the
2
gs boson to invisible final states is measured to be BRi = 0.02 ± 0.20 with
0 from the Zh ! `` +
nnels, while the expected value is 0.00 ± 0.21. If data
-1 -0.8 -0.6 -0.4 -0.2 0
0.2 0.4 0.6 0.8
1
+0.29
BRi
d, the measured (expected) value is BRi = 0.16+0.29
(0.00
0.30
0.32 ) [13]. The
 , BRi ] parametrization used are listed in Models 6 and 7 of Table 1.
Figure boson
6: Likelihood
scanto
of massive
the invisible
branching
BRi is negative because the Higgs
couplings
particles
are ratio of the Higgs boson, BRi , where th
Higgs
measurements:
observed
(55%expected)
@
CL profiled. The observed an
to photons,
, in
andthe
gluons,
g , 95%
have been
e SMSM
values,
so the
measuredloop-induced
overall41%
µh couplings
>1
is accommodated
fit by
likelihoods
are each shown
with
and
without
of expected)
the Zh ! `` +@
ETmiss
channel.
SM
Higgs
measurements+ZH→ll+invisible:
37%
observed
95%
CL Ig
n total
width.
The
smaller expected
uncertainty
when
including
thethe
Zhinclusion
! (39%
physical
boundary BR
the lines atproduces
2 ln ⇤ =an
1.0 and 2 ln ⇤ = 4.0 correspond approx
strates the increase in sensitivity.
Accounting
fori the0,boundary
68% CL (1 ) and 95% CL (2 ), respectively.
CL upper limit of BRi < 0.37 (0.39) using the combination of all channels.
pper limit without including the Zh ! `` + ETmiss data is BRi < 0.41 (0.55).
n the vector boson fusion and associated ZH production modes
CMSのインビジブル崩壊探索
Table 9. Summary of 95% CL upper limits on · B(H !
inv)/ SM obtained from the VBF search, the combined ZH
searches, and the combination of all three searches. arXiv:1404.1344
mH (GeV)
115
125
135
145
200
300
Observed
on
VBF
0.63 (0.48)
0.65 (0.49)
0.67 (0.50)
0.69 (0.51)
0.91 (0.69)
1.31 (1.04)
(expected) upper limits
· B(H ! inv)/ SM
ZH
VBF+ZH
0.76 (0.72) 0.55 (0.41)
0.81 (0.83) 0.58 (0.44)
1.00 (0.88) 0.63 (0.46)
1.10 (0.95) 0.66 (0.47)
—
—
—
—
9 Dark matter interactions
We now interpret the experimental upper limit on B( H !