SM and BSM Higgs Measurements and Searches at ATLAS

Transcripción

SM and BSM Higgs Measurements
and Searches at ATLAS
Elliot Lipeles
University of Pennsylvania
Search 2013
Outline
Diboson Channels ~ Established signals
•Couplings
•Spin-Parity
•Differential Distributions
The Search for Direct Evidence of Fermion Couplings
• H ! ⌧ ⌧, H ! bb, H ! µµ
• ttH production with H !
Higgs Beyond the Standard Model
• Heavy Higgs in H ! W W
• Flavor Changing Neutral Currents in Top Decay
• Higgs to Invisible
Elliot Lipeles
Search 2013
2
The 2011 and 2012 Datasets
Data used in analysis:
-1 7 TeV
2011
5fb
•
-1
•2012 20fb 8 TeV
Elliot Lipeles
Search 2013
3
σ(pp → H+X) [pb]
102
mH =125 GeV
pp →
H(
NN
LO
+NN
LL
10
pp →
1
pp
pp
→
→
pp
ZH
ttH
→
(N
NL
O
O
-1
10
+N
LO
(NNL
OQ
CD +
NLO
(N
NL
O
QC
D
(N
L
QC
D
QC
D
+N
LO
QC
D)
EW
)
g
t
ggF
H
t
t
qqH
W
H
s= 8 TeV
LHC HIGGS XS WG 2012
Higgs Production Summary
+N
EW
)
LO
g
EW)
Large QCD Uncertainties
Sensitive to new physics in
the loop
EW
)
10-2
80 100
200
300
400
1000
MH [GeV]
H
q
ttH
VH
W, Z
q
Access to top coupling
Elliot Lipeles
VBF
(vector
boson
fusion)
W, Z
Usually tagged w/ W/Z decay Small QCD Uncertainties
to leptons (inc. neutrinos)
Distinctive forward jet tags
Search 2013
4
Higgs Decay Summary
mH =125 GeV
Observed:
WW, ZZ, and
Searches:
bb, ⌧ ⌧ , Z , and µµ
Each decay probes a
different Higgs
(Yukawa) coupling
Elliot Lipeles
Search 2013
5
Diboson Channels ~ Established signals
Elliot Lipeles
Search 2013
6
A “Higgs” Boson has been Observed
Higgs: Understanding what have we found
Production
Mechanisms
Decay
Channels
Kinematic
Measurements
Elliot Lipeles
Coupling
Constants
Combined
Fit(s)
Properties/
Quantum
Numbers
Search 2013
7
H!
is high
background, but
resonance allows
background to
be determined
from sideband
H!
Signal/Background ~
1/30
Systematic is largely from the cross-section theory uncertainty
Signal strength relative to SM:
Elliot Lipeles
Search 2013
8
H!
by Production Channel
Diphoton sample divided into
exclusive subsets for different
production mechanisms
80-90% leptonic WH and ZH
remainder ttH
1.9% of Sig
3.4% of Sig
95% of Sig
50% hadronic WH and ZH
remainder ggF
54(76)%VBF for loose(tight)
remainder ggF
75-95% ggF depending on
category
Most categories not very pure in one production mode
ATLAS-CONF-2013-012
Elliot Lipeles
Search 2013
9
H!
Diphoton sample divided into
Result is yieldexclusive
in eachsubsets
category
for different
production mechanisms
80-90% leptonic WH and ZH
remainder ttH
1.9% of Sig
3.4% of Sig
95% of Sig
50% hadronic WH and ZH
remainder ggF
54(76)%VBF for loose(tight)
remainder ggF
75-95% ggF depending on
category
Most categories not very pure in one production mode
ATLAS-CONF-2013-012
Elliot Lipeles
Search 2013
9
H!
H
q
W, Z
q
W, Z
Involved
the
WWH
and
ZZH
couplings
in SM
g
t
H
t
t
ATLAS-CONF-2013-012
Elliot Lipeles
g
Involved
the ttH
coupling
in SM
Search 2013
10
H!
differential cross-sections
ATLAS now has preliminary differential distributions
Procedure
• For each range of
make an m plot pT
• Fit plot for yields using
similar parameterized
backgrounds
• “Unfold” the distribution
to get a particle level result
Elliot Lipeles
Search 2013
11
H!
differential cross-sections
ATLAS now has preliminary differential distributions
Elliot Lipeles
Search 2013
12
H ! ZZ
H ! ZZ ! ```` is
very low background
Signal/Background ~1.5
Main Issue is getting the
highest efficiency without
letting this get out of control
Acceptance:
39% 4µ , 26% 2e2µ/2µ2e ,19% 4e
Elliot Lipeles
Separate VBF and VH
categories added
Search 2013
13
Mass from H !
and H ! ZZ
Some tension between
mass determinations
and ZZ
Main systematics on
mass
Photon energy scale
(from Z ! ee data)
Material modeling
(validated with conversions)
Main systematics on ZZ mass
Electron energy scale
Muon momentum scale
(both validated with J/ data)
Mass combined including systematic correlations
Many detailed cross-checks have been performed
Consistency is at the 2.5 sigma level
Elliot Lipeles
Search 2013
14
H ! WW
H ! W W ! `⌫`⌫ is
intermediate background,
but there is no mass peak
We must model all
backgrounds in detail
Selection is complex and
its effect on signal has to
be modeled
Use approximate
mass variable mT
Signal/Background ~ 1/8
Elliot Lipeles
H to WW relies heavily on theory
Search 2013
15
Review of H to WW Analysis
WW suppression using spin correlations
Roughly at rest
W
!
W
!
H !
Spin=0
Spins have
to add to zero
Elliot Lipeles
W products back-to-back
For each W
spins add to
one
Search 2013
16
Roughly at rest
W
H !
Spin=0
W
!
!
Spins have
to add to zero
Elliot Lipeles
e
+
⌫
µ
⌫¯
For each W
spins add to
one
Search 2013
16
Roughly at rest
W
H !
Spin=0
ConsequencesW
!
!
e
+
⌫
µ
For each W
¯directions
Small angle
ll between charged-lepton ⌫
spins
add
to
have
Small invariant Spins
mass m
ll of the two charged-leptons
one
to add to zero
Elliot Lipeles
Search 2013
16
Consequences of spin correlations
Small invariant mass mll of
the two charged-leptons
Elliot Lipeles
Small angle
ll between
charged-lepton directions
Search 2013
17
Modeling of the WW background
Focusing on 0-jet
•Jet requirements adds
a dependence on the
modeling of QCD jet
emission
•Signal to background
ratio means need to
model WW at the
20% * 1/8 = 2.5% level
to not be dominated
by this uncertainty
Signal
Region
Control
Region Uncertainty on
•We are claiming1.6%
ratio of signal/
modeling!
Use control regions
control is ~ 1.6%
Elliot Lipeles
Search 2013
18
H ! W W sources of uncertainty
From the ATLAS H to WW conference note
µ ⌘ observed / SM
ATLAS-CONF-2013-030
Main systematics are theory uncertainties
Uncertainty is 50% statistical and 50% systematics
Elliot Lipeles
Search 2013
19
Uncertainty on ratio of signal/control is ~ 1.6%
WW cross-section uncertainty ~ 6% using NLO
Example for 10 < mll < 30 GeV
Source
Vary factorization and
renormalization scales
Uncertainty
PDFs
1.5%
Underlying Event and
Parton Shower Models
0.2%
“Modeling”
1.2%
•Choice of Generator
Elliot Lipeles
0.9%
Search 2013
20
H ! W W Categorization plots
0-jet (mostly ggF)
Analysis bins:
•0-jet (ggF), 1-jet(ggF), 2-jet(VBF)
•Same-flavor, opposite-flavor
•2011, 2012
Elliot Lipeles
2-jet (mostly VBF)
Finding H to WW might
show the way to finding
BSM under the SM
Search 2013
21
H ! W W VBF event
Elliot Lipeles
Search 2013
22
Spin
Elliot Lipeles
Search 2013
23
Spin
H!
A spin-1 resonance cannot
decay to two photons
so spin-1 is excluded
Photon spins are not
observed
Spin-2 with initial state of
gg or q q̄ will have different
decay kinematics
⇤
cos ✓ is the angle of photons relative to beam direction
with a correction for the boost of the
system
Selection modified to reduce m
⇤
Elliot Lipeles
cos ✓ correlation
Search 2013
24
H!
Spin
Background
from
sideband
data
The data are fit for signal and
background yields for spin-0 and spin-2
The ratio of the best fit likelihoods is
used as a test statistic to set limits
Only 8 TeV data are used at this point
Elliot Lipeles
Search 2013
25
H!
Spin
Background
from
sideband
data
The data are fit for signal and
background yields for spin-0 and spin-2
The ratio of the best fit likelihoods is
used as a test statistic to set limits
Spin-2 produced by gluon
is excluded
at 99% CL
Only 8 TeV datafusion
are used
at this point
Elliot Lipeles
Search 2013
25
H ! ZZ Spin
Considered Jp: 0+, 0-, 1+,1-,2+
Full kinematics measured = 5 angles
Decay products sensitive to Z spins
BDT trained to distinguish
different spin states
Elliot Lipeles
Search 2013
26
H ! ZZ Spin
Considered Jp: 0+, 0-, 1+,1-,2+
Full kinematics measured = 5 angles
Decay products sensitive to Z spins
BDT trained to distinguish
different spin states
0- is excluded at the 97.8% CL
Elliot Lipeles
Search 2013
26
H ! W W Spin
Discriminating variables: spin-2 looks more like background
Elliot Lipeles
Search 2013
27
H ! W W Spin
2d binned fit
BDT0 vs
BDT2
BDT separating non-Higgs
from spin-0
from spin-1
Elliot Lipeles
Test statistic
likelihood
ratio of spin-0
over spin-2
2-d BDT remapped to 1-d
Fit with Spin-0
Fit with Spin-2
Exclusion of 2+ varies from 99% for
100% q q̄ to 95% for 100% gg production
Search 2013
28
Spin Combination
Spin results from WW, ZZ, and
combined
Jp=2+ excluded at 99.9%
CL independent of fqq̄
Elliot Lipeles
Search 2013
29
Grand Combination
We combine all the inputs
just discussed into global
likelihood fit
Includes correlations
of systematics!
Two sets of results:
•Using preliminary
results from Moriond
including bb̄ and ⌧ ⌧
•Published results using
only , W W , and ZZ
Elliot Lipeles
Preliminary Results with
ggF VBF VH ttH
X
WW X
ZZ X
⌧⌧ X
X
X
X
X
bb̄
X
X
X
X
X
X
Published Results with
ggF VBF VH ttH
X X
WW X X
ZZ X X
X
X
X
Search 2013
30
Inclusion of uncertainties
Each analysis has a table like this....only more complicated
•In order to correctly fit all the data you need to include
these correlations
Elliot Lipeles
Search 2013
31
Evidence of VBF production
3.3𝞂 evidence of
VBF production
Important because if
you’ve only seen ggF
then all measurements
are proportional to
gg
total
Recall we measure things like this
gg
Elliot Lipeles
⇥B
=
gg
⇥
total
Search 2013
32
Coupling Interpretation
Several different models depending on assumptions:
•New particles in loops?
•BSM contributions to total width
(invisible decays, other decays to BSM)?
V
f or g
g
f
f
V
V
t
H
t
t
g
Both combined in 
Elliot Lipeles
Search 2013
33
Relationship between measurements
and coupling
An individual measurement looks like this
gg
⇥B
=
gg
⇥
total
We include this in our fitting
⇥B
gg,SM ⇥ B
gg
=
,SM
2 2
g 
2
H
where we define
total
Elliot Lipeles
/
2
H
Search 2013
34
Example Fit: change couplings to SM
Many combinations of assumptions you can make
Assume only SM particles, but give fermions one
scale factor and bosons another
Elliot Lipeles
Search 2013
35
Example Fit 2: add BSM in loops
Keep SM couplings fixed, but add BSM in loops
Only decays to
SM particles
Include invisible
or other BSM
BRinvisible or non-SM < 0.6 at 95% CL
Elliot Lipeles
Search 2013
36
Scaling to 13 TeV
LHC Plans to Run at 13-14 TeV for the next ~15 years
Have
•2015-2016 13 TeV, 100 fb-1
•2018-2020 13(?) TeV, 300 fb-1
•2011 7 TeV 5 fb-1
-1
-1
2022-202(?)
13(?)
TeV,
3000
fb
2012
8
TeV,
20
fb
•
•
Signal
ggF
VBF
ttH
WH
ZH
qq to WW background
gamma gamma bkg
14 TeV/8 TeV
2.6
2.6
4.7
2.1
2.1
2.1
2.1
In 2016, 4*2.6~10 more signal, 4*2.1~8 more background
Elliot Lipeles
Search 2013
37
H ! Z Search
Another loop induced coupling
Reconstructed in the
+
channel
H!Z !` `
Observed limit is18.2xSM
with 13.5xSM expected
Elliot Lipeles
Search 2013
38
Search for direct evidence of fermion couplings
Elliot Lipeles
Search 2013
39
H ! ⌧ ⌧ Search
Problem is tau decays
involve neutrinos
• ⌧ ! `⌫⌫
• ⌧ ! h⌫ where h is
some number of pions
Large Backgrounds
•Multijet QCD
• Z ! ⌧⌧
Two viable signal
regions:
Boosted ggF
VBF
Make best estimate using
measured constrains of tau mass
and measured missing momentum
Elliot Lipeles
Search 2013
40
H ! ⌧ ⌧ Search
Analysis divided by (tau decay) x (VBF, ggF, Boosted ggF):
•lepton-lepton
•lepton-hadron
•hadron-hadron
Two of the most sensitive categories
Boosted ggF
VBF
Not updated since HCP (Nov 2012)
Elliot Lipeles
Search 2013
41
H ! ⌧ ⌧ Search
Analysis divided by (tau decay) x (VBF, ggF, Boosted ggF):
•lepton-lepton
•lepton-hadron
•hadron-hadron
Two of the most sensitive categories
Boosted ggF
VBF
Upper limit: 1.9xSM (1.2xSM expected)
Significance: 1.1 sigma (1.9 sigma expected)
Signal Stength/SM: µ = 0.7 ± 0.7
Not updated since HCP (Nov 2012)
Elliot Lipeles
Search 2013
41
H ! bb Search
Extremely difficult because
•b-jets are common
•jet resolutions are not so great
•jet distributions hard to model
Highest S/B categories
Abandon gluon fusion (ggF), and
focus on VH production
•V= W ! `⌫, Z ! `` , Z ! ⌫⌫
•Focus on high PT V systems
Very Recent Update = July
Elliot Lipeles
Search 2013
42
H ! bb Search
Highest S/B categories
Fit normalizes backgrounds and exploits
Analysis very complex
different S/B for channels
Signal regions
sliced
based
•Extremely
difficult because • 26 signal
Abandon
gluon
fusion
(ggF), and
regions
with
m
distributions
jj
on
0,
1,
or
2
leptons,
and
b-jets are common
focus on VH production
•based
V bins)
V
(0,1,2
leptons)x(0,1-jet)x(p
T
on
P
T
Z
!
⌫⌫
Z
!
``
W
!
`⌫
resolutions
are
not
so
great
V=
,
,
•jet
•
• 27 control regions
Usedistributions
lower PT V to
control
••jet
hard to model • 26 one
Focus
on+ high
Ptop
systems
T V control
•
eµ
b-tag
1
background modeling
region
Elliot Lipeles
Very
Recent
Update
=
July
After global fit background uncertainty is ~3%
Search 2013
42
H ! bb Search
Even find the
W Z + ZZ ! W bb̄ + Zbb̄
background was hard
Sum over all
channels
weighted by
S/B
WZ+ZZ (to bb) is ~5
times the Higgs signal
Elliot Lipeles
Search 2013
43
H ! µµ Search
If it is really the Higgs it couples to mass,
so H ! µµ should be very small
Limit is 9.8 times SM at
mH = 125 GeV
Predicted signal is under a sea of Drell-Yan
Elliot Lipeles
Search 2013
44
ttH production with H !
Similar to regular H !
but
with either
•Leptonic tag: 1e/µ + 1 b-jet
•Hadronic tag: 6 jets w/ 2 b-jets
Fit m
Elliot Lipeles
just like regular analysis
Search 2013
45
Beyond the Standard Model
Elliot Lipeles
Search 2013
46
Heavy Higgs decaying to WW
Selection similar to SM
but cuts modified for high mass
• pT,lep > 40 GeV,
mll > 50 Ge V, ⌘ll < 1
• still fit in mT
Two widths considered: Narrow
and SM-like
•Couplings unknown means
width unknown
•Interference with gg ! W W
important for wider Higgs
0-jet
(ggF)
2-jet
(VBF)
Analysis includes ggF and
VBF categories like standard
analysis
Elliot Lipeles
Search 2013
47
Heavy Higgs decaying to WW
Selection similar to SM
but cuts modified for high mass
Limits
• pT,lep > 40 GeV,
mll > 50 Ge V, ⌘ll < 1
• still fit in mT
Two widths considered: Narrow
and SM-like
•Couplings unknown means
width unknown
•Interference with gg ! W W
important for wider Higgs
0-jet
(ggF)
2-jet
(VBF)
Analysis
Relative
includes
ggF vs
ggF
VBF
andcross-sections are model dependent
VBF categories like standard
analysis
Elliot Lipeles
Search 2013
47
Flavor Changing Neutral Currents in
Top Decay: t ! cH
Reconstruct H in diphotons
Two mass peaks, one from
each top:
•One mass is mj
•Hadronic reco: 3-jets
•Leptonic reco: mjl⌫
using W mass constraint
4 combinations
per event
Non-resonant background
model is smoothed SHERPA
diphotons + n-partons
Elliot Lipeles
Search 2013
48
Flavor Changing Neutral Currents in
Top Decay: t ! cH
Reconstruct H in diphotons
Two mass peaks, one from
each top:
•One mass is mj
Limit onreco:
top 3-jets
•Hadronic
branching
fraction
mjl⌫
reco:
•Leptonic
0.83%
Observed
using
W mass
constraint
0.53% Expected
4 combinations
per event
Non-resonant background
model is smoothed SHERPA
diphotons + n-partons
Elliot Lipeles
Search 2013
48
Higgs to Invisible
We have also searched directly for Higgs to invisible...
Z recoiling against nothing
SM source of Z recoiling
against nothing
Elliot Lipeles
BRinvisible < 0.65 (observed) at
95% CL, (0.84 expected)
Search 2013
49
Higgs to Invisible Interpretation
Implications for dark
matter searches if
DM to nucleon
couplings is entirely
Higgs
from arXiv:1109.4398v1 [hep-ph]
Based on expected
sensitivity (BRinv<0.75) very
close to observed
Elliot Lipeles
Search 2013
50
Summary
There is an important interplay between theory and
experiment
•Experiment is an input into Theory
•Theory is an input into Experiment
•Some important measurements are close to being
theory limited
Couplings:
•order 20-30% constraints
on vectors, fermions just
crossing the sensitivity
thresholds
•Interesting sensitivity to
dark matter and other BSM
Elliot Lipeles
Spin:
•various combinations of
0-, 1+,1-, and 2+ excluded
Search 2013
51
WW Summary
Elliot Lipeles
Search 2013
52
H!
Elliot Lipeles
VBF BDT
Search 2013
53
H!
Elliot Lipeles
VBF Significance
Search 2013
54
H!
VBF Candidate
VBF Channel has a high purity, S/B ~=
Elliot Lipeles
Search 2013
55
Why bother with H ! W W ?
Electroweak fits make it hard to mess with the
ratio of HWW/HZZ couplings, can’t we just
measure H ! ZZ ?
WW uncertainty is smaller
than ZZ
µW W
µZZ
=
=
1.01 ± 0.31
+0.5
1.7 0.4
Difference will big bigger
at13 TeV, IF the theory
uncertainties can be
controlled
Elliot Lipeles
Search 2013
56
Signature: two e or µ leptons and two neutrinos
Gluon Fuison (ggH)
typically gives 0 or 1 jets
Vector Boson Fusion
gives 2 or more jets
Many of the lessons in H to WW will apply to other searches
where the signal is under significant SM background and does
not peak
Elliot Lipeles
Search 2013
57
Just about everything in the hadron collider zoo
is a background
gg ! H ! W W
Elliot Lipeles
VBF H ! W W
Search 2013
58
W+jets
background
• qq̄ ! W
! l⌫ with
an associated jet...
• the jet is
misidentified as a
lepton
• small background,
but uncertainty is
large
• one of the largest
Hard to predict theoretically, because of
experimental
dependence on fragmentation and
uncertainties
detector response
Elliot Lipeles
Search 2013
59
Z+jets
background
• Different-flavor
(eµ ) background
mainly from Z ! ⌧ ⌧
• Tiny Background
• Same-flavor ( ee & µµ )
background from
⇤
q q̄ ! Z/ ! ll
with false missing
momentum signature
•again, a small
Hard to predict ...
background, but
see next slide
uncertainty is large
Elliot Lipeles
Search 2013
60
Challenges in predicting missing
energy distributions
We make multiple cuts to suppress Z/
⇤
! `` and Z ! ⌧ ⌧
Missing Transverse “Energy”
miss
~
ET =
X
p~T
calorimeter
Missing Transverse Energy Relative
⇢
miss
|E
|
miss,rel
T
ET
⌘
miss
|ET
| sin
if
if
⇡/2
< ⇡/2
miss
~
where
is the angle between ET and the nearest
lepton or jet
miss,rel
ET
Elliot Lipeles
is less sensitive to mismeasurements of leptons and jets
Search 2013
61
energy distributions: Pile-up
•You should think great H ! W W
⇤
! `⌫`⌫ has
miss
ET
and
Z/ ! `` doesn’t, so we are done
•But it’s difficult to measure hadronic energies precisely
•There is still too much left after a reasonable cut for the
same-flavor so we have to use the soft recoil system and
calibrate it with data
•Made much worse by pile-up:
Elliot Lipeles
Search 2013
62
energy distributions: Pile-up
Pile-up is hard to model: Soft
QCD
These properties are very
hard to model
Models are tuned directly to
data, but we still find
modeling issues which grow
with pile-up
Elliot Lipeles
Search 2013
63
energy distributions: Underlying Event
Underlying event is due to a variety of soft QCD effects
•To use simulation, total amount of energy needs to be
simulated in data
•Simulation is tuned to data, but the modeling is limited
Elliot Lipeles
Search 2013
64
Top and
Single top
Looks just like
WW but with
more jets
Single Top
Diagrams
Elliot Lipeles
Search 2013
65
Jet Counting Strategy:
•Divide analysis into
0-jet, 1-jet, and 2+-jets.
•“controls” top
background, but leads
to many of the
uncertainties
Most of the sensitivity
is from 0-jets
Use b-jet “tagging” to
suppress top bkg in1-jet
Elliot Lipeles
Specialize the 2+-jets to
looking for the VBF
signature
Search 2013
66
WW
background
•WW is considered
“irreducible”
•It can be partially
suppressed using the
effect of spin
correlations on the
angles because the
leptons
q q̄ ! W W
Gets
contributions
from
both
•It can also be
gg ! W W
and
suppressed using
gg ! W W will become more important
kinematics
at 13 TeV
Elliot Lipeles
Search 2013
67
Roughly at rest
W
!
W
!
H !
Spin=0
Spins have
to add to zero
Elliot Lipeles
For each W
spins add to
one
Search 2013
68
Roughly at rest
W
H !
Spin=0
W
!
!
Spins have
to add to zero
Elliot Lipeles
e
+
⌫
µ
⌫¯
For each W
spins add to
one
Search 2013
68
Roughly at rest
W
H !
Spin=0
!
e
+
⌫
µ
!
W
For each W
between charged-lepton directions
Consequences
Small angle
ll
⌫
¯
spins
add
to
Small invariant Spins
mass m
of
the
two
charged-leptons
ll
have
one
to add to zero
Elliot Lipeles
Search 2013
68
Consequences of spin correlations
Small invariant mass mll of
the two charged-leptons
Elliot Lipeles
Small angle
ll between
charged-lepton directions
Search 2013
69
WW suppression approximate mass calculation
This obeys the right
basic kinematics
mT < mH
But width of distribution
for both signal and
background is broad
Elliot Lipeles
Search 2013
70
Diboson
Backgrounds
• qq̄ ! W Z/
⇤
! lll⌫
with a lost lepton
⇤
q
q̄
!
Z
! ll⌫⌫ is
•
also a small background
•These are generally modeled with simulation
⇤
⇤
•There is special case W where the is nearly massless
that is difficult to predict
Elliot Lipeles
Search 2013
71
Focusing on 0-jet
•Jet requirements adds
a dependence on the
modeling of QCD jet
emission
•Signal to background
ratio means need to
model WW at the
20% * 1/8 = 2.5% level
to not be dominated
by this uncertainty
Signal
Region
Control
Region Uncertainty on
•We are claiming1.6%
ratio of signal/
modeling!
Use control regions
control is ~ 1.6%
Elliot Lipeles
Search 2013
72
Uncertainty on ratio of signal/control is ~ 1.6%
WW cross-section uncertainty ~ 6% using NLO
Example for 10 < mll < 30 GeV
Source
Vary factorization and
renormalization scales
Uncertainty
PDFs
1.5%
Underlying Event and
Parton Shower Models
0.2%
0.9%
“Modeling”
•Choice of Generator
•Different Generators make difference
1.2%
approximations: zero width, massless bquarks....
Elliot Lipeles
Search 2013
73
Why vary the factorization and renormalization scales?
• An all orders
calculation wouldn’t
depend on these
scales, so any
dependence is a
rough estimate of
the uncalculated
terms
From Michelangelo
Mangano’s slides
Elliot Lipeles
Search 2013
74
Example: ggH signal
•This is actually the single largest systematic uncertainty
on µ ⌘ observed / SM
•I.e the denominator from theory is bigger than all the
experimental errors, but not yet the statistical uncertainty
•We determine the uncertainty from renormalization and
scale variation
Elliot Lipeles
Search 2013
75
Modeling the ggH Signal Acceptance
Next two slides from Stewart-Tackmann,(arXiv:1107.2117 [hep-ph])
lots of strong interactions
• gg ! H is full of gluons
•Jet cuts make it significantly harder to calculate acceptance.
•Adding a jet cut adds a new scale into the problem
Inclusive cross-section
Cross-section requiring a jet pT > 30 GeV
1-jet cross-section looks schematically like this
where
Elliot Lipeles
Search 2013
76
Modeling the ggH Signal Acceptance
Then 0-jet looks like this:
Cancelation between terms
with L and without at
roughly the experimental cut
Suggested procedure to fix
this gives large uncertainty
introduced due to jet cuts
Elliot Lipeles
Search 2013
77
VBF analysis in a slide
The process
Veto jets
between
the tagging
jets in Y
Properties
Keep
Elliot Lipeles
Keep
Search 2013
78
Modeling the VBF Backgrounds
ggF+2 jets
43% uncertainty
from QCD
scale and PDFs
WW+2jets
42% uncertainty from
QCD scale and PDFs
Elliot Lipeles
ttbar
+2jets
15% uncertainty for
extrapolation from
control region
Search 2013
79
H ! W W VBF event
Elliot Lipeles
Search 2013
80
WW Spin
Elliot Lipeles
Search 2013
81
H ! W W Spin
Discriminating variables: spin-2 looks more like background
Elliot Lipeles
Search 2013
82
H ! W W Spin
2d binned fit
BDT0 vs
BDT2
from spin-0
from spin-1
Elliot Lipeles
Test statistic
likelihood
ratio of spin-0
over spin-2
Fit with Spin-0
Fit with Spin-1
Exclusion of 2+ varies from 99% for
100% q q̄ to 95% for 100% ggproduction
Search 2013
83
H ! W W Spin Variables
Elliot Lipeles
Search 2013
84
Spin Combination
Spin results from WW, ZZ, and
combined
Jp=2+ excluded at 99.9% CL independent of
Elliot Lipeles
fqq̄
Search 2013
85
Gamma Gamma VBF
Elliot Lipeles
Search 2013
86
H!
Elliot Lipeles
VBF BDT
Search 2013
87
H!
Elliot Lipeles
VBF Significance
Search 2013
88
H!
VBF Candidate
VBF Channel has a high purity, S/B ~=
Elliot Lipeles
Search 2013
89

SM and BSM Higgs Measurements and Searches at ATLAS

Transcripción

Documentos relacionados

ST. FRANCIS OF ASSISI CATHOLIC CHURCH WHITESBORO