1. Università degli Studi di Torino
Scuola di Dottorato in Fisica
XIX ciclo
Higgs boson detection in vector boson fusion processes by identification of W decays into electrons at CMS detector: a Monte Carlo study
Candidate: Gianluca Petrillo
Turin, Tuesday May 29, 2007

2. Mass in Dirac quantum theory
The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limits
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Mass in Dirac quantum theory
The simplest quantum field theory describes electrodynamics. It's based on the gauge symmetry:
UQ(1)
1 group generator (the electromagnetic charge, Q), 1 vector boson mediating the interaction (the photon);
gauge transformation on new fields:
A'μ(x) = Aμ(x) – e –1 ∂με(x);
Lagrangian is
LD = ψ γμ(i∂μ – e AμQ)ψ – m|ψ|2 + ¼ FμνFμν [ + mA2 AμAμ ];
ψ(x) is a structure representing any single fermion, including its spin degeneration and its antiparticle: the electron (and positron), the electron neutrino (and antineutrino), the proton (and antiproton);
vector field mass mA must be 0 since the last term would not be invariant under gauge transformation.
Turin, May 29, 2007
pp → qq qq eνe
2/50 G. Petrillo

3. Mass in Standard Model SUL(2)×UY(1)×SUC(3)
The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limits
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Mass in Standard Model
The Standard Model is a Yang-Mills field theory, which is based on the symmetry:
SUL(2)×UY(1)×SUC(3)
3+1+8 group generators, vector bosons mediating interaction (3 weak bosons and a photon from mixing of SUL(2)×UY(1) generators, plus 8 gluons);
Lagrangian is
LD = χL γμ(i∂μ + gAμaTa + g'BμY + gSGμaLa) χL + χR γμ(i∂μ + g'BμY + gSGμaLa) χR;
χL(x) are three leptonic «left» fields and three quark fields, χR(x) are six leptonic «right» fields and six quark fields.
the electron of the Dirac theory is now split in two pieces, one sharing the same doublet (χL) with neutrino, and the other in a singlet (χR); the fermion mass term –m|ψ|2 wouldn't be invariant, mixing left and right fields which transform in different ways;
all field masses must be null: gluons, photons, weak bosons, fermions.
Turin, May 29, 2007
pp → qq qq eνe
3/50 G. Petrillo

4. The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limits
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Higgs scalar field
The Standard Model requires the “static” mass terms to be null.
Still, mass can be introduced again as dynamic, thanks to the interaction with a scalar field which has potential:
V(Φ) = λ(Φ†Φ)2 + μ2 Φ†Φ
With the introduction of Higgs field, the Lagrangian includes also:
L = ½ ∂μη∂μη + mW2W+μW–μ + ½ mZ2ZμZμ mW2/v W+μW–μη + mZ2/v ZμZμη + mW2/v2 W+μW–μηη + mZ2/2v2 ZμZμηη – mf f f – mf/v f f η – ½ mη2η2 – ½ mη2/v η3 – ⅛ mη2/v2 η4
the mass terms are reintroduced, but now their non-invariance is reabsorbed by the interaction terms;
the mass of fermions are obtained adding ad hoc Yukawa coupling;
four real fields of Higgs complex doublet Φ(x):
three, ξ(x), disappear from Lagrangian, “absorbed” by weak bosons to get mass;
the fourth, η(x), becomes the physical Higgs boson, acquires mass.
Turin, May 29, 2007
pp → qq qq eνe
4/50 G. Petrillo

5. Vector boson scattering
The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limits
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Vector boson scattering
An important feature of Standard Model, which suggests it as an effective theory, is the cross section of longitudinally polarized vector bosons.
The amplitude for the s-channel of VLVL → VLVL scattering when only weak bosons and Higgs «sectors» are included:
𝐴 𝑚 𝐻 2 𝑣 2 −𝑠+ 𝑠 2 𝑠− 𝑚 𝐻 2 
without Higgs boson, or with mH2 >> s, interaction amplitude is strong and proportional to mH2 s;
with Higgs boson, it correctly vanishes for s >> mH2.
Turin, May 29, 2007
pp → qq qq eνe
5/50 G. Petrillo

6. Higgs boson mass limits
The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limit
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Higgs boson mass limits
The Higgs boson mass is a free parameter of the theory.
Standard Model, including Higgs boson, is considered an effective theory, a low energy limit, valid up to a certain energy scale, of a more general theory
by requiring Higgs boson potential to be bounded and finite up to energy scale Λ, its mass is bounded:
for Λ = 1 TeV scale, 70 GeV/c2 < mH < 750 GeV/c2
for G.U.T. scale (Λ = 1015 GeV): 130 GeV/c2 < mH < 170 GeV/c2
some quantities (exp. masses) are sensitive to Higgs boson loops. The ratio between weak boson masses is sensitive to this: measurements of ρ and mt point to mH < 280 GeV/c2, which is very sensitive to top quark mass value
ρ= 𝑚 𝑊 2 𝑚 𝑍 2 cos 2 θ 𝑊 ≈ π 2 𝑣 2  𝑚 𝑡 2 + 𝑚 𝑊 2 tan 2 θ 𝑊 log 𝑚 𝐻 2 𝑚 𝑊 2 
Turin, May 29, 2007
pp → qq qq eνe
6/50 G. Petrillo

7. Higgs boson search (so far)
The Standard Model
Mass
Higgs field
VV scattering
Higgs mass limit
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Higgs boson search (so far)
Direct search:
LEP: e+e− → Z0H0→ ℓ+ℓ− bb and e+e− → Z0H0→ νℓνℓ bb (LEP1) and e+e− → Z0H0→ f f bb (LEP2)
Tevatron (still running): pp → VH0 (H0 → bb or H0 → W+W−)
from the global «electroweak fit», including data from LEP, Tevatron and others:
mH = 76+33–24 GeV/c2 (CL=68%)
precision measurement: mH < 144 GeV/c2 (CL = 95%)
including LEP2 direct search (mH > GeV/c2): mH < 182 GeV/c2 (CL = 95%)
Turin, May 29, 2007
pp → qq qq eνe
7/50 G. Petrillo
March 2007

8. Large Hadron Collider LHC-b ALICE ATLAS CMS The Standard Model LHC
Features
CMS
Calorimetry
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Large Hadron Collider
LHC-b
ALICE
ATLAS
CMS
Turin, May 29, 2007
pp → qq qq eνe
8/50 G. Petrillo

9. Large Hadron Collider Proton-proton and ion-ion (208Pb) collider
The Standard Model
LHC
Features
CMS
Calorimetry
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Large Hadron Collider
Ebeam
7 TeV
1.7 GeV
Proton-proton and ion-ion (208Pb) collider
40 mb
elastic
12 mb
single diffractive
Startup luminosity (2∙1033 cm−2s−1 = 2 nb/s) → 4.5 inelastic interactions/bunch crossing
Design luminosity (1034 cm−2s−1 = 10 nb/s) → 19 inelastic interactions/bunch crossing
60 mb
inelastic
Turin, May 29, 2007
pp → qq qq eνe
9/50 G. Petrillo

10. The Standard Model
LHC
Features
CMS
Calorimetry
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Compact Muon Solenoid
tracker (pixel detectors and silicon strip detectors): up to 13 3D points
electromagnetic calorimeter (homogeneous, PbWO4 crystals): 26 X0
hadronic calorimeter (sampling, argon/brass and quartz/lead): 7–10 λI
solenoidal magnet (superconducting, niobium-titanium): 4 T (inner), 2 T (muons)
muon detector (drift tubes, cathode strip chambers, resistive plate chambers)
Turin, May 29, 2007
pp → qq qq eνe
10/50 G. Petrillo

11. CMS calorimetry ECAL (barrel): ∆η = 0.0175; ∆φ = 0.0175 rad
The Standard Model
LHC
Features
CMS
Calorimetry
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
CMS calorimetry HF
(η = 5)
ECAL (barrel):
∆η = ; ∆φ = rad
HCAL (barrel):
∆η = ; ∆φ = rad
Turin, May 29, 2007
pp → qq qq eνe
11/50 G. Petrillo
σ 𝐸 𝐸 = 2.7% 𝐸 ⊕ 210𝑀𝑒𝑉 𝐸 ⊕0.55%
σ 𝐸 𝐸 = 110% 𝐸 ⊕4.5%

12. Higgs boson production at LHC
The Standard Model
LHC
Higgs boson
Production
Decay
VV features
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Higgs boson production at LHC
Turin, May 29, 2007
pp → qq qq eνe
12/50 G. Petrillo

13. Higgs boson decay Higgs coupling is proportional to:
The Standard Model
LHC
Higgs boson
Production
Decay
VV features
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Higgs boson decay
Higgs coupling is proportional to:
mass of the fermion
squared mass of the boson
The search channels:
H0 → γ γ (the CMS “golden channel”), either gg or VV fusion
H0 → τ+τ− in V-V fusion, very hard to tag
H0 → Z Z* → ℓ+ℓ−ℓ+ℓ−, the “golden plated” channel
H0 → t t, high multiplicity channel, very hard.
Turin, May 29, 2007
pp → qq qq eνe
13/50 G. Petrillo

14. H0 → VV reconstruction features
The Standard Model
LHC
Higgs boson
Production
Decay
VV features
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
H0 → VV reconstruction features
If mH > 150 GeV/c2, the most likely decay channel is into two weak bosons, from which many final states are available.
Higgs boson branching ratio (BR) similar into both W W and Z Z;
at least one weak boson must decay into leptons
branching ratio less than halved
fully leptonic
BR: 10% (W → eνe, W → μ νμ) and 3% (Z0 → e+e−, Z0 → μ+μ−)
leptonic W decay (WL)
neutrino involved, not easy to “detect”, can't be fully reconstructed; WW channel decaying into four leptons can't reconstruct the Higgs mass (just its transverse mass)
hadronic V decay (VH)
prone to be faked by jets from other physical processes
V-V fusion production channel
“spectator” quarks, useful in tagging but can feed the misreconstruction of VH
electrons are slightly easier to be faked by jets than muons.
Turin, May 29, 2007
pp → qq qq eνe
14/50 G. Petrillo

15. pp → qq → qq H0 → qq eνe @CMS q't,1 e W q1 νe V q2 qV q'V q't,2
The Standard Model
LHC
Higgs boson
Processes
V-V fusion
Backgrounds
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
pp → qq → qq H0 → qq
This analysis: Higgs by vector boson fusion into W bosons and finally into two quarks, an electron and its neutrino (BR = 7.3%).
The “roles”: q1 q2
q't,1
q't,2 qV e νe
q'V ? V W
leptons (WL boson)
expected central in η, from a W decay hopefully coming from Higgs boson
“central” quarks (VH boson)
expected more central in η, from decay of the other W or Z from Higgs boson
“tag” quarks
correlated with incoming quarks and protons, expected at higher η, one forward, the other backward
This analysis won't exploit the whole CMS detector capabilities, but it will concentrate on calorimetry only, bypassing the information from muon chambers and tracker.
Turin, May 29, 2007
pp → qq qq eνe
15/50 G. Petrillo

16. QCD background processes
The Standard Model
LHC
Higgs boson
Processes
V-V fusion
Backgrounds
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
QCD background processes
Same final state: qq qq eνe: 4 jets, an electron, missing energy, always with W → eνe.
pp → tt → bb W+W− (α4ewα2s)
pp → VW qq (α4ewα2s)
pp → W qqqq (α2ewα4s)
pp → tb qq → bb qq W (α4ewα2s)
Similar final state:
pp → Z qqqq, Z → e+e−, missing energy is faked (α2ewα4s)
pp → ZZ qq, ZZ → qq e+e−, missing energy is faked (α4ewα2s)
For each of them, additional QCD jets, soft and not-that-soft, can be added.
There is chance of faking both electron and missing energy with jets; Z can provide missing energy as well.
Turin, May 29, 2007
pp → qq qq eνe
16/50 G. Petrillo

17. Software PHANTOM events ALPGEN events FAMOS reconstructed events
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Software
PHANTOM events
ALPGEN events
FAMOS...
FAMOS reconstructed events
generated objects
reconstructed objects
generation
Tagging
Combinations...
roles recognition
χ2-like quantities
roles assignment
reconstruction
reconstruction information
goodness of roles assignment
Turin, May 29, 2007
pp → qq qq eνe
17/50 G. Petrillo
analysis

18. Events generation: PHANTOM
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Events generation: PHANTOM
PHANTOM (was PHASE) generates events with 6 fermions in final state including processes with order α6ew and α4ewα2s and using exact matrix element calculation.
qq → qq qq eνe q1 q2
q't,1
q't,2 qV e νe
q'V q1
q'1 V W
γ,W±,Z0
q'2 q2 q1 q2 W V V H0 W- W+ q1
q'1 V W
q'2 q2 V
γ,V,H0 W W- V W+
Turin, May 29, 2007
pp → qq qq eνe
18/50 G. Petrillo

19. Events generation: PHANTOM (II)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Events generation: PHANTOM (II)
The official PHANTOM production has been used (data taking time is in design luminosity scenario, 100 fb−1/year):
Generation parameters:
Turin, May 29, 2007
pp → qq qq eνe
19/50 G. Petrillo

20. Events generation: ALPGEN
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Events generation: ALPGEN
ALPGEN is a set of Monte Carlo generators devoted to hadron collider processes.
It employs matrix element calculation, includes polarization effects in many processes with weak bosons and supports jet matching to avoid double counting.
The following samples have been generated:
Turin, May 29, 2007
pp → qq qq eνe
20/50 G. Petrillo

21. Signal sample extraction (I)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Signal sample extraction (I)
Events in PHANTOM samples include all α6ew diagram contributes. Which ones are Higgs-resonant?
diagrams separation is conceptually wrong: all diagrams interfere!
I need a “signal” sample as benchmark for cuts
some event categories showing no Higgs boson peak can be defined
I use these categories as an effective criterion to isolate signal at generation level q1 q2
q't,1
q't,2 W Z q1
q'1 V W
γ,W±,Z0
q'2 q2 b W+ W− t q1 q2 W V q1 q2
q't,1
q't,2 W
Turin, May 29, 2007
pp → qq qq eνe
21/50 G. Petrillo

22. Signal sample extraction (II)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Signal sample extraction (II)
Event categories:
processes which must go through annihilation instead of vector boson fusion: identified by quark flavour. «untagged»
processes with a top quark inside: identified by b quark plus W invariant mass. «topped»
processes with 3 vector bosons in final state: identified with tag quarks invariant mass. «3V»
processes which can't resonate in Higgs boson anyway: identified by vector bosons charge.
the remaining: they are compatible with vector boson fusion, have no top or 3V dominant component, have a 2W final state. «Higgs signal» (even if no Higgs is generated!) q1
q'1 V W
γ,W±,Z0
q'2 q2 b W+ W− t q1 q2 W V q1 q2
q't,1
q't,2 W Z q1 q2
q't,1
q't,2 W
Turin, May 29, 2007
pp → qq qq eνe
22/50 G. Petrillo

23. Signal sample extraction (III)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Signal sample extraction (III) q1
q'1 V W
γ,W±,Z0
q'2 q2 b W+ W− t
«topped»
«Higgs» q1 q2
q't,1
q't,2 W q1 q2 W V
«untagged»
«3V»
Turin, May 29, 2007
pp → qq qq eνe
23/50 G. Petrillo

24. Generated WW mass The Standard Model LHC Higgs boson Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Generated WW mass
Turin, May 29, 2007
pp → qq qq eνe
24/50 G. Petrillo

25. Samples summary The Standard Model LHC Higgs boson Processes Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Samples summary
Turin, May 29, 2007
pp → qq qq eνe
25/50 G. Petrillo

26. Fast Simulation (FAMOS)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Fast Simulation (FAMOS)
The CMS fast detector simulation and reconstruction, FAMOS
contains a simplified CMS detector geometry
adds minimum bias events to the generated one
propagates the particles from scattering through detector and magnetic field
simulates the interaction of particles with detector material: bremsstrahlung, photon conversion, multiple scattering...
generates detector “hits” and responses
applies reconstruction algorithms to compund higher level objects
supports only partially the reconstruction algorithms available in full reconstruction software
doesn't simulate trigger response
has pile-up been tuned only for startup luminosity (2∙1033 cm−2s−1)
is badly documented
is now deprecated and discontinued, new software is available in place
Turin, May 29, 2007
pp → qq qq eνe
26/50 G. Petrillo

27. Reconstructed objects
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstructed objects
To reconstruct this event, they are needed: q1 q2
q't,1
q't,2 qV e νe
q'V
electron
missing energy
four jets
Turin, May 29, 2007
pp → qq qq eνe
27/50 G. Petrillo

28. Reconstruction: electrons
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstruction: electrons
FAMOS reconstructs electrons in different steps:
calorimeter hits (crystals of ECAL with deposited energy)
basic clusters: group of hits with energy possibly deposited by the same particle (photon, electron, pion... with two algorithms)
clusters: basic clusters with the most proper algorithm of the two
superclusters: collections of clusters along φ direction; supposed to group the cluster from an electron with the photons it emitted
electron and a photon (matching the superclusters with tracker)
In this analysis, tracker data have not been used. The electron candidate for the vector boson fusion channel will be extracted among supercluster. This provides:
energy (E)
position of the hit (η, φ): this includes the magnetic field effect
no E/p information, since p would be from tracker
no charge information (e+/e−/γ)
no H/E nor isolation, which are not provided but can be computed
Turin, May 29, 2007
pp → qq qq eνe
28/50 G. Petrillo

29. The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstruction: jets
FAMOS reconstructs jets by different algorithms and calibrations:
start from “elements”: calorimeter hits or tracker trajectories)
grouped by k┴ algorithm or iterative cone, which is:
compile a sorted list of seeds from the most energetic elements;
for each seed, assign all elements within a ΔR = (Δη2+Δφ2)½ < 0.5 radius as belonging to a proto-jet
compute proto-jet center as average of positions of its elements, weighted by energy;
iterate again by a new cone ΔR < 0.5 around the new center
keep iterating until the result is stable
mark components of the final proto-jet as used, go for next seed
elements recombination scheme:
jet axis: weighted average of positions of elements
transverse energy: vector sum of E┴ of elements (“E┴ scheme”)
energy calibration: just one is available, which is simulated by γ- jet events.
Turin, May 29, 2007
pp → qq qq eνe
29/50 G. Petrillo

30. Reconstruction: missing energy
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstruction: missing energy
FAMOS missing ennergy reconstruction is not extremely refined:
uses calorimeter towers, slices of ECAL crystals and HCAL tiles along a common direction
energy of a calorimeter tower is just the sum of the ECAL and HCAL components, no intercalibration is used
total “scalar” transverse energy (SET): sum of all transverse energies of calorimeter towers
total missing transverse energy (MET): opposite to the vector sum of all transverse energies of calorimeter towers, with the axis of their tower as direction
In this analysis, I use the missing transverse energy as a reconstruction of the transverse energy of the sole neutrino.
Turin, May 29, 2007
pp → qq qq eνe
30/50 G. Petrillo

31. Leptons selection (WL): neutrino
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Leptons selection (WL): neutrino
Missing energy:
just one “candidate” from reconstructed missing energy
required to have MET > 30 GeV/c2 and MET/SET > 7%
Turin, May 29, 2007
pp → qq qq eνe
31/50 G. Petrillo

32. Leptons selection (WL): electron
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Leptons selection (WL): electron
Electron:
candidates from ECAL superclusters
required E┴ > 15 GeV
required to be isolated in a ∆R < 0.3 respect to jets, accepted even if, instead, the only jet in cone has Ejet < 2 Esclus;
Turin, May 29, 2007
pp → qq qq eνe
32/50 G. Petrillo

33. Leptons selection: WL For each neutrino-electron candidate pair:
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Leptons selection: WL
For each neutrino-electron candidate pair:
transverse momentum is computed by weak boson mass constraint
the resulting mass must be lighter than 110 GeV/c2 range and the transverse mass inside [40 – 105] GeV/c2 range
the pair with the most likely transverse mass is chosen; if none has survived, the event is discarded
Turin, May 29, 2007
pp → qq qq eνe
33/50 G. Petrillo

34. Generated quark distributions
(left) Quarks from WH decay have the same distribution in both signal and irreducible background samples from PHANTOM. Distribution of tag quarks is very different instead: in signal tag jets are very separated, which results in a larger ∆η.
(right) Transverse momentum of all generated quarks from all samples are plotted. Quarks from W+jets backgrounds are weaker than the ones from PHANTOM samples.
Turin, May 29, 2007
pp → qq qq eνe
34/50 G. Petrillo

35. Jets selection (WH, tags)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Jets selection (WH, tags)
Four must be picked among reconstructed jets to identify tag and central jets. Candidate jet layouts are chosen by these criteria:
only the 5 jets with the highest p┴ are considered;
tag jets: p┴ > 30 GeV/c, mj > 10 GeV/c2, ηFW > –1, ηBW < +1
most central jet: p┴ > 25 GeV/c, mj > 5 GeV/c2, |η| < 2
other central jet: p┴ > 20 GeV/c, mj > 5 GeV/c2, |η| < 2.5
mtags > 200 GeV/c2, mcentral є [ 60 ; 110 ] GeV/c2
Central jets are chosen first, as the pair whose mass is closest to W one; then tag jets are chosen as the ones with highest invariant mass.
Event is then required to have |Δηtags| > 3, |Δηc| < 1 and central jets between tag ones in η, or it's discarded.
Turin, May 29, 2007
pp → qq qq eνe
35/50 G. Petrillo

36. Jets selection: summary
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Jets selection: summary
Turin, May 29, 2007
pp → qq qq eνe
36/50 G. Petrillo

37. Event selection: summary
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Event selection: summary
The selections on leptons and jets are independent.
The following table shows the events which have passed both.
All these events have a completely reconstructed pair of tag quarks, one lepton-decaying weak boson and a hadronic decaying one, and have a Higgs boson candidate.
Turin, May 29, 2007
pp → qq qq eνe
37/50 G. Petrillo

38. VW mass: resolution The Standard Model LHC Higgs boson Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
VW mass: resolution
Turin, May 29, 2007
pp → qq qq eνe
38/50 G. Petrillo

39. Hard process energy mWWjj > 1.5 TeV/c2
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Hard process energy
Vector boson fusion processes have larger cross section in high s regions respect to QCD background processes. This is expecially true in top events, but when tt has at least an additional jet, that is likely to be labelled as tag jet. This increases the total energy of the six fermions.
Hard process energy is estimated as the sum of 4-momenta of the reconstructed tag jets and V bosons.
mWWjj > 1.5 TeV/c2
Turin, May 29, 2007
pp → qq qq eνe
39/50 G. Petrillo

40. Minimum W-jet mass min{mWj} > 150 GeV/c2
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Minimum W-jet mass
As the vector bosons in signal tend to be central and the tag jets to be forward/backward peaked, the invariant mass of any of the two V bosons plus any of the two tag jets is large: a cut on the minimum of the 4 combinations helps rejecting backgrounds, even if this feature has already in part been exploited.
The cut could be tighter, but it would deplete specifically a range of VW invariant mass plot, biasing the distribution.
min{mWj} > 150 GeV/c2
Turin, May 29, 2007
pp → qq qq eνe
40/50 G. Petrillo

41. Top events mtop in 120 ― 220 GeV/c2
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Top events
Similar to the previous, the cut aimed to suppress top events by identifying top mass events where any jet layout candidate presented a VH + tag jet combination with mass in top mass window: mWj.є [ 120 ; 220 ] GeV/c2.
mtop in 120 ― 220 GeV/c2
Turin, May 29, 2007
pp → qq qq eνe
41/50 G. Petrillo

42. Additional jets E┴jjjj/E┴total > 70%
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Additional jets
To cut events with more than 4 energetic jets, the four labelled jets are required to have a large fraction of the total (scalar) transverse energy from jets.
This has a deep impact on my tt + jets sample, but worsen the signal/W + jets ratio.
E┴jjjj/E┴total > 70%
Turin, May 29, 2007
pp → qq qq eνe
42/50 G. Petrillo

43. The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Final events summary
Trying to maximize purity of reconstruction has caused low efficiency.
Background cross section is still hundred times the signal sample.
Turin, May 29, 2007
pp → qq qq eνe
43/50 G. Petrillo

44. Reconstructed VW mass (I)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstructed VW mass (I)
Turin, May 29, 2007
pp → qq qq eνe
44/50 G. Petrillo

45. Reconstructed VW mass (II)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Reconstructed VW mass (II)
Turin, May 29, 2007
pp → qq qq eνe
45/50 G. Petrillo

46. Higgs peak in no Higgs scenario
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Higgs peak in no Higgs scenario
𝑑σ 𝑑𝑚 Δ𝑚= 𝐴 1+  𝑚 𝑉𝑊 − 𝑚 𝐻 Γ 𝐻 2   𝑚 𝑉𝑊 𝑚 𝐻  α
Signal + background fit:
εσ=𝐿 π 2 Γ Δ𝑚 𝑃
L = 100 fb-1
Fitted peak significance:
peak/σpeak should be null...
Turin, May 29, 2007
pp → qq qq eνe
46/50 G. Petrillo

47. Background smoothing ×3.9 ×207 ×4.1 ×25 𝑑σ 𝑑𝑚 Δ𝑚=  𝑚 𝑚 0  α
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Background smoothing
𝑑σ 𝑑𝑚 Δ𝑚=  𝑚 𝑚 0  α
×3.9
×207
×4.1
×25
Turin, May 29, 2007
pp → qq qq eνe
47/50 G. Petrillo

48. Upper limit: no Higgs scenario
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Upper limit: no Higgs scenario
Use smooth background + PHANTOM sample
Assume the sample coming from real data: Poisson errors!
Search for the peak height limit which would not be considred signal
CL=95%
L = 100 fb-1
100 evt ↔ 71 fb
(ΓH = 50 GeV)
Turin, May 29, 2007
pp → qq qq eνe
48/50 G. Petrillo

49. Peak upper limit (300/500 GeV/c2)
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal
Backgrounds
Signal extraction
Reconstruction
Superclusters
Jets
MET
Leptons
Complete event
Cuts
mWWjj
min{mWj}
top
total ET
mVW
Statistical analysis
Peak fit
Upper limits
Conclusions
Peak upper limit (300/500 GeV/c2)
[mH = 300 GeV/c2]
CL=95%
L = 100 fb-1
1 fb/50 GeV/c2 ↔ 65 fb
(ΓH = 50 GeV)
[mH = 500 GeV/c2]
CL=95%
L = 100 fb-1
1 fb/50 GeV/c2 ↔ 150 fb
(ΓH = 100 GeV)
Turin, May 29, 2007
pp → qq qq eνe
49/50 G. Petrillo
Differences on upper limit are tiny!

50. The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Summary
the analysis of vector boson fusion scattering process qq → qq qq eνe at CMS has been performed using exact signal computation
the analysis has been performed using only CMS calorimetry
the presence of QCD backgrounds is overwhelming
this analysis is affected by low statistics due to the huge amount of background events that would be needed
background shape is important in trying to fit the mass distribution
never the less, interesting limits to Higgs boson cross section can be obtained with this analysis
Turin, May 29, 2007
pp → qq qq eνe
50/50 G. Petrillo

51. Backup slides

52. Interaction and radiation lengths
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Backup
Interaction and radiation lengths
Turin, May 29, 2007
pp → qq qq eνe
backup G. Petrillo

53. Reconstructed WL resolution
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Backup
Reconstructed WL resolution
Turin, May 29, 2007
pp → qq qq eνe
backup G. Petrillo

54. Synthetic QCD background
The Standard Model
LHC
Higgs boson
Processes
Analysis
Software chain
Events generation
Signal extraction
Reconstruction
Leptons
Jets
Complete event
Cuts
mVW
Statistical analysis
Conclusions
Backup
Synthetic QCD background
Weighted sum of the four QCD background samples separately fitted with a power function.
Errors by propagation of the ones of fit parameters.
Turin, May 29, 2007
pp → qq qq eνe
backup G. Petrillo

55. PYTHIA and generated initial state
Hard processes are inserted in a proton-proton scattering process by PYTHIA, which “dresses” both initial and final states. Proton remnants are added, charged particles can emit photons, quarks can emit gluons and finally hadronize. Hadrons decay, except for some kaons, pions and nucleons.
PDF used by PHANTOM and ALPGEN, CTEQ5LO, have quark contributions up to strange quark. The heavier quark in initial states are produced by a partonic gluon. This is how the bb initial state is achieved
Turin, May 29, 2007
pp → qq qq eνe
(backup) G. Petrillo

56. Electron reconstruction: resolution
The distance in ΔR between the direction of the generated electron and the closest supercluster is representative of spatial resolution of electron reconstruction.
Energy resolution is computed as the relative error of energy of reconstructed supercluster respect to the one of the generated electron, when these are less than ΔR = 0.5 far from each other.
Both the resolutions are not affected by selection cuts.
Turin, May 29, 2007
pp → qq qq eνe
(backup) G. Petrillo

57. Neutrino reconstruction: resolution
(left) relation between generated neutrino and reconstructed missing transverse energy; for ΔR, the recognition of the electron is required
(below) resolution on transverse missing energy as measurement of νe transverse energy
Turin, May 29, 2007
pp → qq qq eνe
(backup) G. Petrillo