Standard Model Higgs at LHC

1
Standard Model Higgs at LHC Majid Hashemi For the
CMS and ATLAS collaborations University of
Antwerp, Belgium
2
LHC has 4 main detectors
CMS Size 21 m long, 15 m wide and 15 m high.
Weight 12 500 tonnes Location Cessy, France.
ATLAS Size 46 m long, 25 m high and 25 m wide.
Weight 7000 tonnes Location Meyrin,
Switzerland.
ALICE Size 26 m long, 16 m high, 16 m wide
Weight 10 000 tonnes Location St
Genis-Pouilly, France
LHCb Size 21m long, 10m high and 13m wide
Weight 5600 tonnes Location Ferney-Voltaire,
France.
3
Current status of SM Higgs Searches

• Results already obtained from LEP (in the same
  tunnel before upgrading to LHC) shows a Higgs
  boson mass lower limit of 114.4GeV at 95 C.L.
  (CERN-EP 2003-011)
• Indirect searches including fits to data from
  electroweak measurements set an upper limit of
  182GeV. (hep-ex 0710.4983)
• 114.4 GeV lt m(H) lt 182
  GeV

New

• Tevatron already excluded the mass range between
  160 and 170 GeV !
• arXiv 0903 4001 hep-ex
• Including recent results on W mass measurements
  in Tevatron an indirect upper limit of
  m(H)lt157GeV is obtained from EW fits.

New

• Conclusion
• 114.4 GeV ltm(H)lt157 GeV

4
CMS and ATLAS Higgs search channels

• ATLAS search channels

• CMS search channels

• H??? and H?tt are used in both experiments for
  low mass Higgs boson search m(H)lt135 GeV

• H?WW and H?ZZ are used for intermediate and high
  mass searches 130ltm(H)lt700 GeV.

5
CMS
6
H?gg
7

• CMS has done two
• analyses for H-gtgg search

• Cut based analysis
• A good chance of discovery in low mass
  region at 30fb-1

Signal significance and exclusion limits
8
2) MVA analysis (neural net) Cut on
NN output and then look at m(gg)
tight cut on NN output NNgt0.97
Loose cut on NN output NNgt0.85
9
5-15fb-1 data needed for a 5s discovery in CMS
Mass measurement at the level of 100-200 MeV
possible in CMS
10
ttH, H?bb
11

• ttH in CMS Working point 60fb-1
• Studies three final states
• fully hadronic ttH -gt bbbbqqqq
• fully leptonic ttH -gt bbbblnulnu
• Semi-leptonicttH -gt bbbbqqlnu (le or mu)

12
Loose cuts
Tight cuts

• Systematic uncertainties included in top plots,
• Upper(lower) bound is optimistic(pessimistic)
  bound with signal up(down) and background
  down(up) by 10 and 20 respectively.
• The point with dB(Xsec)0 corresponds to blue
  point.
• The red star is the most optimistic point with
  dB(Xsec) ttNj1 and dB(Xsec) ttbb4

CMS PTDR It is interesting to note that it does
not quite yield a substantial significance, even
though background uncertainties of 1 and 4 for
ttNj and ttbb are probably substantially better
than what will be accessible in reality. This
highlights the challenge that is faced in
observing ttH.
13
qqH , H?tt
14

• t identification crucial for this search,
• Electron rejection requirement
• Hottest hcal tower should have a low
• energy deposit in case of electrons,

• CJV and TCV are also compared as a tool to reject
  Zjets background

15
di-t mass reconstruction including backgrounds
di-t mass reconstruction with different mass
hypotheses
Di-tau Mass measurement resolution
16
Sources of systematic uncertainties for signal
(left) and backgrounds (right)
In CMS with early data of 1fb-1 an upper limit on
the cross section can be established
17
H?ZZ
18
Higgs boson candidate mass at preselection level
Selection cuts
19
Z boson mass reconstruction after final selection
cuts are applied
Higgs boson candidate mass reconstruction after
final selection cuts
20
Systematic uncertainties small and dominated by
statistical uncertainty
Number of events and significances for different
m(H)
A 95C.L. exclusion is possible at early data of
1fb-1 in CMS for the mass range above 180 GeV
21
H?W W
22
1)Cut-based analysis

• Kinematic distributions of signal and background
  events
• A low lepton pair ?f and invariant mass is
  expected for the signal events compared to
  backgrounds,

• After all selection cuts there are 31 events of
  emu signal and 31 background at 1fb-1

Masses lower or higher than m(H)160GeV leave
less signal and more background.
23
2) Multivariate analysis
The neural net analysis gives better results than
cut based analysis
At 1fb-1 with m(H)160GeV, Cut based analysis
S/B70/70, NN analysis S/B67/37
24

• The main background samples are controlled using
  different strategies
• ttbar events, Control region definition
• For ttbar events a control region is introduced
  and number of such events is estimated in the
  signal region using the standard formula
• Control region is close to the signal region
• Selection cuts are basically the same dropping
  central jet veto
• The error of the estimation of the background is
  then calculated using

gt many of the systematics cancel.
18
Vary jet Et by /-7 (jet energy scale
uncertainty)
Statistical uncertainty of observed events
Fluctuation of the background in normalization
region
Reff(CJV in signal region)/eff(2jets in
normalization region)
25

• WW background, Control region definition
• WW events are enhanced with the same selections
  as in ttbar but keeping CJV,
• Optimization is also applied to increase the WW
  sample size in the normalization region,
• The total error including statistical uncertainty
  and backgroud fluctuation is 22.
• Wjet background, fake lepton study
• Fake muons estimate define the probability of a
  loosely isolated track to pass muon id,
• Fake electron estimate define the probability of
  a jet to pass electron id,
• Run this analysis on QCD events ( plenty of
  events) and obtain fake rates,
• Re-weight signal search which is looking for
  loosely isolated tracks and jets,
• The final error estimate is better because a
  large sample of QCD is used for fake rate
  measurement

26
Including all systematic uncertainties the signal
significance is calculated for the cut-based and
neural net analysis
Certainly the multivariate analysis is performing
better and for the central region a 5sigma
discovery is possible.
27
A Higgs boson in the range of 140ltm(H)lt200 GeV
can be excluded at 95C.L. with the data
collected at 1fb-1.
28
ATLAS
29
H?gg
30

• ATLAS searches in different categories of
• Inclusive H-gtgg, or with 1 jet, 2jets and/or Met
• Mass distribution results shown in plots.

Signal and background samples used
Higgs boson candidate mass
inclusive
1jet
2jets
Met
31

• Different analysis categories and number of
  signal and background in each

• Signal significance is calculated with number
  counting and fitting approaches.
• The 1D fit is performed to get the di-photon
  mass.

Results in terms of signal significance A 4s
signal is observable at 10fb-1
32
H?ZZ
33

• H?ZZ samples used in ATLAS

Signal samples

• The irreducible ZZ background is the main
  challenge

Background samples

• Event selection chain
• Online e,m Trigger
• Offline
• Pre-selection,
• kinematic cuts,
• isolation

34

• Higgs boson Mass resolution obtained in ATLAS for
  different final states

4e
2e2m
4m

• Mass resolutions as a function of the true Higgs
  boson mass

4e
4m
2GeV mass reconstruction accuracy is possible at
30fb-1
35
Higgs boson candidate lying on top of background
distributions
M(H)130GeV
M(H)150GeV
M(H)300GeV
M(H)180GeV
The closer to ZZ invariant mass (180GeV),

the harder discrimination between signal
and background!
36
Signal significance at 30fb-1

• Ratio of data needed for exclusion at 95 C.L. to
  data already collected at 5fb-1
• Points below 1 could be excluded at 5 fb-1
  at 95CL.

37
qqH , H-gttt
38

• qqH is favored against gg fusion process to
  exploit the forward-backward jet tagging as a
  signature of signal events
• Leptons and taus are identified with the
  following requirements

tau-jets identification efficiency and fake-tau
rate
39
h distributions of hardest jet and second hardest
jet

• Signal Signature
• h(j1) , h(j2)
• Jets accompanying the Higgs boson tend to have
  large h
• D(hjj) and M(jj)
• Jets go back-to- back resulting in
  large D(hjj) and M(jj)

D(hjj) and M(jj) distributions of hardest jet and
second hardest jet
40

• Signal significance in different channels as a
  function of m(H)
• Results obtained for 30fb-1.

• Linearity of reconstructed mass as a function of
  input mass having subtracted the residual error

Rec. m(H) Gen. m(H) GeV
41
SB B

• Fit to pseudo-data for different m(H)
• With both signalbackground and background-only
  hypotheses.

• Exclusion is possible at 95CL at 10fb-1 almost
  for all analyzed masses with lh channel
• H-gttautau-gtl jet nn

42
H?W W
43
Signal and background samples used for H-gtWW
study

• One of the main discriminating tools for the
  signal is central jet veto as there should be no
  central jet in the event.
• This is a rejection tool against ttbar events.

44

• H0jet -gt WW -gtenmn
• 3 dimentional fit to the following variables

Mass measurement as a function of the input mass
5 GeV error
Variation of Pull of m(H) with systematic
uncertainties not significant
45

• VBF H-gtWW-gtenmn
• Forward backward jets as discriminating tools
  against backgrounds

• Two analyses for VBF search with H-gtWW (-gtllnn)
• 2D fit on Neural network output and Higgs boson
  mT
• (neural net training has to be done before
  the fit)
• 5D fit on mT,?f(ll), ??(ll), ??(jj),m(jj) (less
  model dependent)

46
VBF H-gtWW (-gtllnn)
2D fit result 5D fit result
VBF H-gtWW (-gtlnjj)
VBF H-gtWW (-gtllnn)
47
H/0j, H-gtWW (ll)
H/2j, H-gtWW (ll)
H/2j, H-gtWW (ll)
2D fit, NN output mT
5D fit
The best mass measurement (combination of all
channels)
At 10fb-1 all masses have the discovery chance in
ATLAS
48
ttH,WH, H-gtbb
49

• ttH and WH with H-gtbb analyses have also been
  carried out in ATLAS

ttH (H-gtbb) shape in cut based analysis
30fb-1 LHC data

• ATLAS conclusion
• No clear signal is observed
• Background uncertainties hard to control
• More simulated data is needed
• The analysis is very challenging!

50

• Conclusions
• CMS and ATLAS are performing very well for Higgs
  boson searches,
• CMS is ready for early data, there are several
  analyses just ready to digest the first data,
• ATLAS is more focusing on fits to signal shapes
  and mass reconstruction with data of few fb-1,
• With the early data there is a possibility of
  confirming current Tevatron results,
• a mass range of 140GeVltm(H)lt200GeV can be
  excluded at 95C.L.
• The first discovery is foreseen to be possible
  after 1fb-1 collected data, starting from
  m(H)160GeV and going to lower masses with more
  data,
• The Higgs boson mass resolution is estimated to
  be at the level of 100-200 MeV for low masses
  with H-gtgg at 30fb-1 and 2-3GeV with H-gtZZ at
  30fb-1