Discovery potential for T ′ → tZ
in the trilepton channel at the LHC
Lorenzo Basso
IPHC, Strasbourg
Based on LB, J. Andrea, JHEP 1502 (2015) 032
arXiv:1411.7587 [hep-ph].
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
1 / 27
Outline
Simplified model for T ′ analyses
Analysis
- Simulation details
- Cut-based
- Multi-variate (BDT)
Results
- Discovery potential
- Comparison to dilepton channel
Reinterpretation for top anomalous couplings
Conclusions
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
2 / 27
Introduction
√
s = 7/8 TeV: found the Higgs boson
√
LHC run-II at s = 13(14) TeV: (among others) study Higgs boson properties,
clarify if SM boson, search for New Physics
LHC run-I at
Which NP? To stay with the scalar sector, many BSM theories predicts new
heavy fermions to stabilise the Higgs boson mass and to protect it from
dangerous quadratic divergences
Generally, heavy partners of the third generation quarks with vector-like
couplings: Extra Dimensions, Little Higgs Models, Composite Higgs Models
They are vector-like because their LH and RH chiralities couple to the gauge
boson vectors in the same way
→ LH and RH couplings are equal, as for a vector
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
3 / 27
Examples of models
Very quick review:
J. Reuter and M. Tonini, JHEP 1501 (2015) 088 [arXiv:1409.6962]
Composite Higgs models: Higgs boson is a composite state
tR ∼ 14 , complete rep. of SO(4)
Minimal case: SO(5)/SO(4)
qL ∼ incomplete rep. of SO(5)
14 : T ′
New fermions: Ψ
44 : (T ′ , B ′ ), (X5/3 , X2/3 )
Little Higgs models: Higgs is a pseudo-Goldstone boson
from a global spontaneous breaking of SU (5)/SO(5) (Littlest Higgs model)
A vector-like heavy top is required to cancel loop quadratic divergences
Many models, many similarities → simplified model
Here, singlet top partner: T ′
Typically, BR(T ′ → qW ± ) : BR(T ′ → qZ) : BR(T ′ → qh) ∼ 2 : 1 : 1
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
4 / 27
Simplified model
M. Buchkremer et al. Nucl.Phys. B876, 376 (2013) [1305.4172]
LT ′ = g ∗
q
nq
RL √g
+ µ
′
1+RL 2 [T L Wµ γ dL ]
g
RL
1+RL 2 cos θW
[T ′ L Zµ γ µ uL ] +
q
+
q
1
+ µ
′
√g
1+RL 2 [T L Wµ γ bL ]+
g
1
1+RL 2 cos θW
o
[T ′ L Zµ γ µ tL ] + h.c.
We allow for generic mixing to 1st generation quarks
Only 3 parameters:
- MT ′ , the vector-like mass of the top partner
- g ∗ , the coupling strength to SM quarks, only relevant in single production.
σ ∝ (g ∗ )2
- RL , the mixing coupling to first generation quarks. RL = 0 corresponds to
coupling to t/b only
This talk: single
decay mode via T ′ → tZ,
√ top partner production, trilepton
−1
at the LHC at s = 13 TeV and L = 100 fb
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
5 / 27
Single production and T ′ → tZ
σpp→T ′ (MT ′ , RL )
=
BRT ′ →tZ (MT ′ , RL )
=
MT ′ (GeV)
800
1000
1200
1400
1600
Lorenzo Basso (IPHC)
RL
1
+ A3 (MT ′ )
1 + RL
1 + RL
1
B(MT ′ )
1 + RL
A1 (MT ′ )
A1 (MT ′ ) (pb)
1.2614
0.7752
0.5001
0.3331
0.2265
A3 (MT ′ ) (pb)
0.07242
0.03518
0.01826
0.00994
0.00561
TPP Seminar
B(MT ′ ) (%)
22.4
23.5
24.0
24.2
24.4
March 04, 2015
6 / 27
Monte Carlo simulation details
LO samples simulation with
parton level: MG5_aMC@NLO (CTEQ6L1)
Hadronisation/showering: Pythia6 Tune Z2
FastSim: Delphes3 ma5Tune
Analysis: MadAnalysis5
Signal:
5 benchmark points of T ′ mass in steps of 200 GeV: MT ′ ∈ [800; 1600] GeV,
with g ∗ = 0.1 and RL = 0.5. No k-factors
Backgrounds (plus up to 2 jets):
- 3 prompt leptons: ttW , ttZ, tZj, and W Z
- non-prompt leptons: tt and Z/W + jets
Samples normalised to NLO cross sections where available
CMS detector emulation
Anti−kT algorithm with R = 0.5
b-tagging CVS medium working point: b−tag = 70%, mistag = 1%
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
7 / 27
Courtesy of Eric Conte
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
8 / 27
More simulation details
Massive background event generation to gather enough statistics:
Process
SingleTop_W_madspin
SingleTop_t_5FS_madspin
TTdilep_WWToLLNuNu_madspin
TTdilep_ZToLL_madspin
TTdilep_madspin
TTsemilep_WToLNu_madspin_2
TTsemilep_WWToLLNuNu_madspin_2
TTsemilep_ZToLL_madspin_1
TTsemilep_ZZToLLLL_madspin_1
TTsemilep_madspin_1
TZq2_W_trilep1
TZq2_s_trilep
WToLNu-0Jet_sm-no_masses
WToLNu-1Jet_sm-no_masses
WToLNu-3Jets_sm-no_masses
WZToLLJJ
WZToLNuNuNu
ZToLL10-50-0Jet_sm-no_masses
ZToLL10-50-2Jets_sm-no_masses
ZToLL50-0Jet_sm-no_masses
ZToLL50-2Jets_sm-no_masses
ZToLL50-4Jets_sm-no_masses_split
ZZToLLLL
# Files
189
83
1
1
200
1
1
1
1
172
100
94
592
586
488
5
1
1
1
9
9
1
92
# Events
18898481
8299246
99999
99989
9427953
59771
99997
99995
99993
8105465
9999157
9393276
52785449
32827404
12931463
306339
59147
97701
38998
784399
350088
2884
6222800
Process
SingleTop_s_madspin
TTdilep_WToLNu_madspin
TTdilep_WZToLLLNu_madspin
TTdilep_ZZToLLLL_madspin
TTsemilep_WToLNu_madspin_1
TTsemilep_WWToLLNuNu_madspin_1
TTsemilep_WZToLLLNu_madspin_1
TTsemilep_ZToLL_madspin_2
TTsemilep_ZZToLLLL_madspin_2
TTsemilep_madspin_2
TZq2_W_trilep2
TZq2_t5FS_trilep
WToLNu-0Jet_sm-no_masses-run2
WToLNu-2Jets_sm-no_masses
WWToLLNuNu
WZToLLLNu
WZToNuNuJJ
ZToLL10-50-1Jet_sm-no_masses
ZToLL10-50-3Jets_sm-no_masses
ZToLL50-1Jet_sm-no_masses
ZToLL50-3Jets_sm-no_masses_split
ZZTo4Nu
ZZToLLNuNu
# Files
188
1
1
1
1
1
2
1
1
173
97
97
482
396
194
120
1
1
1
10
8
1
1
# Events
18771372
64191
99991
99993
59694
99989
199988
99987
99990
8156688
9672987
9699081
42972689
15769022
11221071
7666801
59420
45361
5690
549567
115396
35808
64305
Monte Carlo errors below permil: neglected
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
9 / 27
Objects selection
Objects identification
pT (ℓ) > 20 GeV,
pT (j) > 40 GeV,
|η(j)| < 5,
Background
tt(+X)
tZj
WZ
Total
MT ′ (GeV)
800
1000
1200
1400
1600
no cuts
7.5 106 (100%)
3521 (100%)
1.4 105 (100%)
7.6 106 (100%)
no cuts
119.7 (100%)
77.1 (100%)
52.0 (100%)
35.3 (100%)
24.5 (100%)
|η(e/µ)| < 2.5/2.4 ,
∆R(ℓ, j) > 0.4 ,
(1)
(2)
|η(b)| < 2.4,
(3)
1 ≤ nj ≤ 3
6.1 106 (81.2%)
2953 (83.9%)
5.7 104 (41.9%)
6.1 106 (80.5%)
1 ≤ nj ≤ 3
105.0 (87.8%)
67.8 (87.9%)
45.3 (87.2%)
30.5 (86.6%)
21.1 (86.0%)
nℓ ≡ 3
514.9 (0.09%)
290.6 (9.8%)
3883 (6.9%)
4689 (0.08%)
nℓ ≡ 3
39.3 (37.4%)
26.0 (38.4%)
16.1 (35.6%)
8.0 (26.1%)
3.8 (18.0%)
nb ≡ 1
243.8 (47.3%)
170.0 (58.5%)
164.3 (4.2%)
578.0 (12.3%)
nb ≡ 1
25.5 (64.8%)
16.4 (63.2%)
10.1 (62.4%)
4.8 (60.1%)
2.2 (58.3%)
Signal generated without taus
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
10 / 27
Cut-based analysis: optimisation
Z-boson reco by minimising distance of OSSF leptons to MZ
s = 13 TeV
3
10
2
10
10
WZ
TT (+X)
TZj
MT’ = 0.8 TeV
MT’ = 1.0 TeV
MT’ = 1.2 TeV
MT’ = 1.4 TeV
MT’ = 1.6 TeV
1
-1
10
-2
10
-3
10
10-4
0
50
100
150
200
250 + 300
M(l l ) (GeV)
|M (ℓ+ ℓ− )/ GeV − MZ | < 15
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
11 / 27
Cut-based analysis: optimisation
W reco with remaining lepton
s = 13 TeV
10
1
WZ
TT (+X)
TZj
MT’ = 0.8 TeV
MT’ = 1.0 TeV
MT’ = 1.2 TeV
MT’ = 1.4 TeV
MT’ = 1.6 TeV
-1
10
-2
10
-3
10
-4
10
0
50
100
150
200
250
300
MT (lW ) (GeV)
10 < MT (ℓW )/GeV < 150
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
11 / 27
Cut-based analysis: optimisation
top reco with remaining lepton and b-tagged jet
10
1
-1
10
-2
10
WZ
TT (+X)
TZj
MT’ = 0.8 TeV
MT’ = 1.0 TeV
MT’ = 1.2 TeV
MT’ = 1.4 TeV
MT’ = 1.6 TeV
-3
10
-4
10
0
50
100
150
s = 13 TeV
200
250
300
MT (blW ) (GeV)
0 < MT (ℓW b)/GeV < 220
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
11 / 27
Cut-based analysis
Selections
|M (ℓ+ ℓ− )/ GeV − MZ |
10 < MT (ℓW )/GeV
0 < MT (ℓW b)/GeV
Background
tt(+X)
tZj
WZ
Total
MT ′ (GeV)
800
1000
1200
1400
1600
nb ≡ 1
243.8 (47.3%)
170.0 (58.5%)
164.3 (4.2%)
578.0 (12.3%)
nb ≡ 1
25.5 (64.8%)
16.4 (63.2%)
10.1 (62.4%)
4.8 (60.1%)
2.2 (58.3%)
cut (4)
154.8 (63.5%)
155.6 (67.2%)
146.9 (89.4%)
457.2 (79.1%)
cut (4)
23.8 (93.6%)
15.4 (93.8%)
9.5 (94.2%)
4.5 (93.5%)
2.1 (93.3%)
< 15 ,
(4)
< 150 ,
< 220 .
(5)
(6)
cut (5)
135.1 (87.3%)
148.7 (95.6%)
138.2 (94.1%)
422.0 (92.3%)
cut (5)
22.2 (93.2%)
14.3 (92.4%)
8.7 (92.3%)
4.1 (92.1%)
1.9 (92.2%)
cut (6)
83.0 (61.5%)
139.8 (63.7%)
71.5 (51.7%)
294.3 (69.8%)
cut (6)
20.8 (93.6%)
13.4 (94.0%)
8.1 (92.3%)
3.8 (91.3%)
1.7 (90.0%)
Cuts optimised to retain ≥ 90% of signal
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
12 / 27
Cutflow
7
10
s = 13 TeV
6
10
5
10
4
10
TTsLL
TTdLL
TT
TZq
TTLNu
WZ
Tp_800
Tp_1000
Tp_1200
Tp_1400
Tp_1600
3
10
2
10
10
1
no c
uts
0<N
Lorenzo Basso (IPHC)
j<4.
lep=
3
nbje
ts =
|Mll
M Tl
-MZ M Tl<1
b<2
50
|<15
20
1
TPP Seminar
March 04, 2015
13 / 27
Events (L=100 fb-1)
MT (b 3ℓ)
s = 13 TeV
10
WZ
TT (+X)
TZj
MT’ = 0.8 TeV
MT’ = 1.0 TeV
MT’ = 1.2 TeV
MT’ = 1.4 TeV
MT’ = 1.6 TeV
1
10-1
-2
10
200
400
600
800
1000
1200
1400 1600 1800
MT (b3l) (GeV)
Signal clearly visible over background
Distribution in transverse mass, sharper peaks than invariant mass
Q: can we do better?
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
14 / 27
Multi-Variate Analysis (MVA)
Cut-based strategy: suitable cuts on the most straightforward distributions
Is it the best strategy?
Many additional variables to distinguish signal from background
Recall the kinematics: T ′ very heavy, t − Z back-to-back and boosted
However, cutting on any of these variables unavoidably reduce also the signal
Solution:
combine several variables using a multivariate analysis (MVA) to obtain the
best signal/background discrimination
Here we used Boosted Decision Tree (BDT)
Variables drawn after Z mass cut: MT (ℓW ), MT (ℓW b); MET, HT , ST ; pT , η;
∆η, ∆φ, angular correlations, . . .
Some variables correlated, like pT (Z) and pT (ℓ1 ): choose a reduced and
uncorrelated set with still large sensitivity
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
15 / 27
0.22
0.2
0.18
0.12
0.08
0.12
0.1
0.06
0.08
0.1
0.04
0.05
0.02
1000
1500
0.045
0.04
1
1.5
2
2.5
3
3.5
4
4.5
5
∆ R(b l )
0.3
0.25
0.2
0.03
0.15
0.02
0.1
0.015
0.01
0.05
0.005
0
2
0
4 max
η (j)
0.16
1
2
3
4
5
6
∆ φ (ll )
0.4
0.6
0.8
0.1
0.08
0.08
0.06
0
1
1.2
1.4
p (Z)/M (b3l) (GeV)
T
-6
-4
-2
0
2
4
T
0.14
6
∆ φ (t Z)
Background
0.12
MT’ = 1.0 TeV
0.1
0.08
0.06
0.04
0.02
1
2
3
4
5
0
6
∆ φ (Z MET)
0.1
0.2
0.4
0.6
0.8
1
1.2
1.4
p (j )/M (b3l) (GeV)
T 1
T
0.1
0.08
0.1
0.12
0.2
Z
0.12
0.14
0.05
0.24
0.22
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
1/N dN
-2
1/N dN
-4
1/N dN
0
0.04
0.02
0
W
0.035
0.025
0.1
0.06
0
0.5
2000
2500
MT(b3l) (GeV)
1/N dN
500
1/N dN
0
0.2
0.15
1/N dN
0.2
0.3
0.25
0.16
0.14
0.1
0.15
1/N dN
0.16
0.14
1/N dN
MT’ = 1.0 TeV
0.25
1/N dN
Background
0.3
1/N dN
0.35
1/N dN
1/N dN
MVA variables
0.08
0.06
0.06
0.06
0.04
0.04
0.04
0.04
0
-3
-2
-1
0
1
2
3
∆ η (ll )
Z
0
0.02
0.02
0.02
0.02
-4
-3
-2
-1
0
1
2
3
4
∆ η (b l )
0
-6
W
-4
-2
0
2
4
6
8
η (top)
0
1
2
3
4
5
6
∆ φ (Z l )
W
+ ηZ
pT (Z)/MT (b 3ℓ), pT (j1 )/MT (b 3ℓ), and MT (b 3ℓ) are decorrelated
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
16 / 27
MVA variables II
Variable
MT (b 3ℓ)
pT (Z)/MT (b 3ℓ)
η max (j)
∆ϕ(Z, p
/T )
∆η(ℓℓ|Z )
η(t)
η(Z)
Importance
2.60 10−1
9.41 10−2
6.02 10−2
5.37 10−2
5.05 10−2
4.99 10−2
4.61 10−2
Variable
∆R(b, ℓW )
∆ϕ(t, Z)
∆ϕ(ℓℓ|Z )
pT (j1 )/MT (b 3ℓ)
∆η(b, ℓW )
∆ϕ(Z, ℓW )
Importance
9.77 10−2
8.17 10−2
5.89 10−2
5.08 10−2
5.03 10−2
4.63 10−2
(ℓℓ|Z ): the pair of leptons reconstructing the Z boson
η max (j): jet with largest rapidity (to account for associated jet)
pT (j1 )/MT (b 3ℓ) and pT (Z)/MT (b 3ℓ) effectively decorrelated from MT (b 3ℓ)
Angular variables from fully reconstructing the neutrino 4-momentum
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
17 / 27
Correlations - MT ′ = 1 TeV
Correlation Matrix (signal)
Linear correlation coefficients in %
tree_MtZ
-1
tree_hdphitZ
tree_dRblW
tree_detablW
-2
1 -17
-2
-9
6
1
1
-1
1
1 -29
8
7
tree_dphiZlW
-5
tree_dphiZMET
1 -31
tree_DetaZll
-4
21
1 100
-4
-2 -17 100
1 100 -17
tree_ptZratio
-2
7
tree_etaZ
18
-1
tree_etaJhrec -10
tree_ptJ1ratio
100
100
tree_etaTop 100
6 100
1
3 100
6
12
-1
-9
1
-4
-1
11 13
-29
2
7
1
6
-1
7
-2
-31 -5
-4
-1
1
-2
40
20
-20
-40
1
-60
-2
-10 18
60
0
-1
8
-2
4
1
1
1
-15
3
100
80
1 -25
100
11
-1
tree_dphiZll
100
1
100
100
1 100
-1
13
2
-1 12 -25
4 -15
-17
21
1
-1
-80
-100
tree tree tree tree tree tree tree tree tree tree tree tree tree
_eta _ptJ _eta _eta _ptZ _dp _De _dp _dp _de _dR _hd _Mt
Top 1rat Jhre Z
ratio hiZll taZll hiZM hiZlWtablW blW phitZ Z
io
ET
c
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
18 / 27
BDT output
(1/N) dN / dx
TMVA overtraining check for classifier: BDT
6
Signal (test sample)
Signal (training sample)
Background (test sample)
Background (training sample)
Kolmogorov-Smirnov test: signal (background) probability = 0.921 (0.336)
5
U/O-flow (S,B): (0.0, 0.0)% / (0.0, 0.0)%
4
3
2
1
0
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
BDT response
Allows to check for “overtraining”: 2 random samples, one used for training
and the other one for comparison, should get similar output
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
19 / 27
Discovery power: benchmarks
Surviving events and significances for signal benchmark points
(g ∗ = 0.1, RL = 0.5)
C&C: select a window around the peak in MT (b3ℓ)
MVA: perform a LH cut on BDT output
√
to maximise the significance: σ = S/ S + B
Analysis
MT (b3ℓ) cut (GeV)
S (ev.)
C&C
B (ev.)
σ
cut
MVA
σ
MT ′ = 0.8 TeV
[800 − 860]
18.00
8.90
3.47
0.07
3.64
MT ′ = 1.0 TeV
[840 − 1200]
12.28
4.88
2.96
0.08
3.10
MT ′ = 1.2 TeV
[1000 − 1340]
7.16
1.74
2.40
0.11
2.50
MT ′ = 1.4 TeV
[1120 − 1640]
3.40
0.90
1.64
0.12
1.62
MT ′ = 1.6 TeV
[1200 − 1800]
1.57
0.63
1.06
0.12
1.15
MVA: non-significant improvement (5%–8%)
Significance depends on g ∗ and RL per fixed T ′ mass
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
20 / 27
Discovery power: parameter space
g* vs M
g* vs R
L
g*
g*
T’
0.5
0.45
0.5
0.45
0.4
0.4
0.35
0.35
0.3
0.3
0.25
0.25
0.2
0.2
0.15
0.15
RL=0 (0%)
0.1
0.1
RL=1.0 (50%)
0.05
0
800
0.05
RL=3.0 (75%)
RL=9.0 (90%)
1000
1200
1400
1600
1800
2000 2200
MT’ (GeV)
0
0
1
2
3
4
5
6
MTp=1.6 TeV
MTp=1.4 TeV
MTp=1.2 TeV
MTp=1.0 TeV
MTp=800 GeV
7
8
9
RL
(dashed lines: 5σ, solid lines: 3σ)
T ′ masses up to 2 TeV can be observed
Increased reach when RL is non-vanishing (maximum for RL ≃ 1,
corresponding to 50%–50% mixing)
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
21 / 27
Comparison to dilepton channel
We set ourselves in similar conditions: L = 300 fb−1 , κf = 1.14, RL = 0
g* vs M
g*
T’
J. Reuter and M. Tonini,
0.5
JHEP 1501 (2015) 088 [arXiv:1409.6962]
0.45
0.4
0.35
0.3
0.25
0.2
0.15
RL=0 (0%)
0.1
RL=1.0 (50%)
0.05
0
800
RL=3.0 (75%)
RL=9.0 (10%)
1000
1200
1400
1600
1800
2000 2200
MT’ (GeV)
(dashed lines: 5σ, solid lines: 3σ)
Comparable reach at low T ′ masses (no pair-prod. here)
200 ÷ 300 GeV better sensitivity at high T ′ masses
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
22 / 27
Reinterpretation: top anomalous couplings
The top-quark couplings can be parametrised in an effective field theory
The SM Lagrangian is extended by gauge-invariant (non-renormalisable)
operators, obtained by integrating out heavy modes
L = LSM +
X Ci Oi
i
Λ2
Here we consider only dimension 6 operators, the first non-vanishing terms in
1/Λ expansion: total of 59 operators W. Buchmuller, D. Wyler, Nucl.Phys. B268 (1986) 621
Not all possible dim-6 operators that one can write are independent
Redundant operators can be reduced by using equation of motions and other
relations due to gauge invariance
J. A. Aguilar-Saavedra, Nucl. Phys. B812, 181 (2009) [0811.3842]
L=
g κtZq µν L
R
√
tσ
fZq PL + fZq
PR q Zµν ,
Λ
2cW
q=u,c
X
where Λ is the scale of new physics.
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
23 / 27
Reinterpretation: optimisation
tZq coupling gives similar final state as T ′ → tZ → t ℓ+ ℓ−
s = 13 TeV
3
10
102
WZ
TT (+X)
TZj
K zct -BR=1%
K zut-lim
K γ ut -lim
10
1
-1
10
-2
10
-3
10
10-4
0
50
100
150
200
250 + 300
M(l l ) (GeV)
tγq coupling (with γ ∗ ) too. However, the cut around MZ removes it
Here, couplings at best limit: κtZ(γ)q -lim= 0.2(0.1) TeV−1
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
24 / 27
Events (L=100 fb-1)
Reinterpretation: C&C
s = 13 TeV
10
WZ
TT (+X)
TZj
K zct -BR=1%
K zut-lim
1
-1
10
Cut
no cuts
1 ≤ nj ≤ 3
nℓ ≡ 3
nb ≡ 1
cut (4)
cut (5)
cut (6)
MT (b3ℓ)
S
B
σ
κtZu
2263(100%)
1765(78.0%)
191.8(10.9%)
113.8(59.3%)
103.2(90.7%)
96.2(93.3%)
91.1(94.7%)
> 400 GeV
68.0
102.9
5.2
κtZc
5360(100%)
4452(83.0%)
623.3(14.0%)
381.0(61.1%)
342.7(90.0%)
323.6(94.4%)
304.7(94.1%)
> 200 GeV
304.5
241.7
13.0
-2
10
200
400
600
800
1000
1200
1400 1600 1800
MT (b3l) [GeV]
MVA trained on T ′ signals and applied to top anomalous couplings does not
lead to improvements
In progress: training on the top anomalous signal
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
25 / 27
Reinterpretation: parameter space
BR(t->Zc) (%)
σ
BRs
-1
9
L = 300 fb
8
L = 100 fb-1
0.3
10
0.25
8
7
0.2
6
6
5
0.15
5σ
4
3
4
3σ
0.1
2
2
0.05
2σ
1
0
0.05
0.1
0.15
0.2
0.25
KtZu (TeV-1)
0.01
0.02
0.03
0.04
0.05
0.06
0.07
BR(t->Zu) (%)
Actual limits: BR(t → Zq) < 0.05% (inclusive, from tt) ⇒ κtZu < 0.2 TeV−1
BR(t → Zu) < 0.51%
Otherwise from single top:
BR(t → Zc) < 11.4%
See CMS-TOP-12-037 and CMS-TOP-12-021, respectively
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
26 / 27
Conclusions
Singlet top partners T ′ common to many BSM models trying to address the
Higgs stability
Simplified model: only 3 parameters, simple rescaling to cover whole phase
space
T ′ → tZ: study of the trilepton signature at
mode
√
s = 13 TeV in single production
T ′ masses up to 2.0 TeV and couplings down to g ∗ = 0.1 can be probed.
Large gain if mixing with light generation is accounted for
Results from cut-based analysis: simple and effective, no substantial
improvements from MVA
Reinterpretation to top anomalous couplings sharing similar signature
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
27 / 27
Backup slides
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
28 / 27
∆R(ℓ+ ℓ− ) for T ′ signals
WZ
TT (+X)
TZj
MT’ = 0.8 TeV
MT’ = 1.0 TeV
MT’ = 1.2 TeV
MT’ = 1.4 TeV
MT’ = 1.6 TeV
102
10
1
-1
10
-2
10
-3
10
-4
10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
∆ R(ll) (GeV)
Z
T ′ is very massive, hence the decay products are boosted
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
29 / 27
Correlations - Background
Correlation Matrix (background)
Linear correlation coefficients in %
tree_MtZ
16 -29
-11
tree_hdphitZ
7
-19
6
5 -18
10
100
-7
100
-11
7
7
-2 100
-18
7
tree_dphiZMET
-22
-14
4
100 -2
5
4
-29
2
-42 100
4
7
10
tree_ptZratio
10
100 -42
-14
7
-21
tree_etaZ
21
39 100
10
tree_etaJhrec
23
100 39
2
100
tree_etaTop 100
10
23
2
20
0
100
tree_dphiZll
tree_ptJ1ratio
60
40
tree_dphiZlW
tree_DetaZll
100
80
100
-21 10
2
100
-7
1
tree_dRblW
tree_detablW
4
6
-20
16
-40
-60
-19
-22 -11
21
1
-11
-80
-100
tree tree tree tree tree tree tree tree tree tree tree tree tree
_eta _ptJ _eta _eta _ptZ _dp _De _dp _dp _de _dR _hd _Mt
Top 1rat Jhre Z
ratio hiZll taZll hiZM hiZlWtablW blW phitZ Z
io
ET
c
Lorenzo Basso (IPHC)
TPP Seminar
March 04, 2015
30 / 27