The Road to Discovery · 2018-11-21 · Gustaaf Brooijmans CERN 2017 Goals Searches span very wide...
Transcript of The Road to Discovery · 2018-11-21 · Gustaaf Brooijmans CERN 2017 Goals Searches span very wide...
Gustaaf Brooijmans CERN 2017
The Road to Discovery Searching for BSM Physics at Hadron Colliders
Gustaaf Brooijmans
1
Gustaaf Brooijmans CERN 2017
Goals❖ Searches span very wide spectrum - Overview of many types of searches, with focus on experimental aspects
- Challenges posed at hadron colliders
- Ease of generating false positives
- Techniques to deal with limited knowledges
- But far from exhaustive!
❖ Thread in results - Hints?
2
Gustaaf Brooijmans CERN 2017
Standard Model Today
3
Triumph of Gauge Theories!
Gustaaf Brooijmans CERN 2017
Standard Model Today❖ Higgs discovery completes the
Standard Model - Fully consistent, complete, precise
description of strong, electromagnetic and weak interactions
❖ Even generate fermion masses - But that is the only property of
fermions we “understand”
4
Gustaaf Brooijmans CERN 2017
In Words❖ Matter is built of spin 1/2 particles that interact by exchanging 3 different
kinds of spin 1 particles corresponding to 3 different (gauge) interactions
❖ There appear to be 3 generations of matter particles
❖ The 4 different matter particles in each generation carry different combinations of quantized charges characterizing their couplings to the interaction bosons
❖ The matter fermions and the weak bosons have “mass”
❖ Gravitation is presumably mediated by spin 2 gravitons
❖ Gravitation is extremely weak for typical particle masses
❖ There appear to be 3 macroscopic space dimensions
5
Gustaaf Brooijmans CERN 2017
About the Standard Model❖ It’s a theory of interactions: - Properties of fermions are inputs
- In gauge paradigm, fermion properties “generate” interactions
- Properties of interaction bosons in terms of couplings, propagations, masses are linked:
- Measuring a few allows us to predict the rest, then measure and compare with expectation
❖ It’s remarkably successful: - Predictions verified to be correct at sometimes incredible levels of precision
- After ~40 years, still no serious cracks
6
Gustaaf Brooijmans CERN 2017
Precision Results
7
muon g-2: 0.7 ppm!
B, K physicsLEP, SLD & Tevatron
Gustaaf Brooijmans CERN 2017
Lacking in the Standard Model❖ Clear structure in fermionic sector
unexplained - No understanding of the “charges”
- Evidence of selective principle(s) - E.g. no neutral colored fermions
- q(down) = q(e)/Nc
- Interpreted as evidence for (grand) unification
- Grand or less grand? (One or more scales?)
8
Gustaaf Brooijmans CERN 2017
Lacking in the Standard Model❖ Many cosmological issues - Dark matter and dark energy
- Not enough CP violation in the quark sector for baryogenesis
- Baryon number violation - Present in the SM through B-L (sphalerons)
- Baryogenesis through leptogenesis and B-L?
‣ Untestable?
9
Gustaaf Brooijmans CERN 2017
Many Fundamental Questions❖ What exactly is spin? Or color? Or electric charge? Why are they
quantized? ❖ Are there only 3 generations? If so, why? ❖ Why are there e.g. no neutral, colored fermions? ❖ What is mass? Why are particles so light? ❖ Is there a link between particle and nucleon masses? ❖ How does all of this reconcile with gravitation? How many space-time
dimensions are there really? ❖ ...
10
Gustaaf Brooijmans CERN 2017
Particles Solve Problems
(Problems Predict Particles)
11
Gustaaf Brooijmans CERN 2017
Vector Boson Scattering
12
• There was in fact one known problem with the Standard Model (+ a second, related, lesser one):
• If we collide W’s or Z’s (not so easy...), the scattering cross-section grows with the center of mass energy, and gets out of control (violates unitarity) at about 1.7 TeV: σ(WW → WW) ∼ s
• This is similar to “low” energy neutrino scattering:
• If q2 << (MW)2, looks like a “contact interaction”, and cross-section grows with center of mass energy: σ ∼ s
• But when q2 ≈ (MW)2, W-boson propagation becomes visible, and “cures” this problem
Gustaaf Brooijmans CERN 2017
Vector Boson Scattering
13
• There was in fact one known problem with the Standard Model (+ a second, related, lesser one):
• If we collide W’s or Z’s (not so easy...), the scattering cross-section grows with the center of mass energy, and gets out of control (violates unitarity) at about 1.7 TeV: σ(WW → WW) ∼ s
• This is similar to “low” energy neutrino scattering:
• If q2 << (MW)2, looks like a “contact interaction”, and cross-section grows with center of mass energy: σ ∼ s
• But when q2 ≈ (MW)2, W-boson propagation becomes visible, and “cures” this problem
Gustaaf Brooijmans CERN 2017
The Higgs Boson❖ One way to solve WW, is to introduce a massive, spinless particle (of
mass < ~1 TeV) - Couplings to W and Z are fixed, quantum numbers are known...
- .... to be those of the vacuum
- Its mass is unknown, and its couplings to the fermions are unknown.... well, maybe
- Fermions can acquire mass by coupling to this Higgs boson, so their couplings could be proportional to their masses. This is called the “Standard Model Higgs”
14
Gustaaf Brooijmans CERN 2017
Precision Measurements❖ In fact, we were able to say
something about the standard model Higgs mass
- If the fermions get their masses from the Higgs, we know all couplings and can infer the Higgs mass from precision measurements
- Result is very sensitive to measured top quark, W boson masses
- Really wants a “light” Higgs boson15
Gustaaf Brooijmans CERN 2017
Precision Measurements❖ In fact, we were able to say
something about the standard model Higgs mass
- If the fermions get their masses from the Higgs, we know all couplings and can infer the Higgs mass from precision measurements
- Result is very sensitive to measured top quark, W boson masses
- Really wants a “light” Higgs boson16
✓
Gustaaf Brooijmans CERN 2017
The Plot Thickens
17
Gustaaf Brooijmans CERN 2017
New Physics?
18
Could this be it?
Gustaaf Brooijmans CERN 2017
Higgs Mass❖ Higgs, in fact, also acquires
mass from coupling to W’s, fermions, and itself!
- These “mass terms” are quadratically divergent
- Drive mass to limit of validity of the theory
❖ So we expect the Higgs mass to be close to the scale where new physics comes in....
19
Gustaaf Brooijmans CERN 2017
New Physics?
20
Could this be it?
Gustaaf Brooijmans CERN 2017
New Physics?
21
Could this be it?
Gustaaf Brooijmans CERN 2017
Nevertheless❖ Clear structure in fermionic sector
unexplained - Evidence of some selective principle
(why are there no neutral colored fermions?)
- Proton stability, running of couplings suggestive of at least one other scale relevant to SM particles and interactions, ~1015 GeV
- Either fine-tuning, or a closer scale
22
Gustaaf Brooijmans CERN 2017
The Tools
23
Gustaaf Brooijmans CERN 2017
Energy Frontier❖ Currently, hadron colliders: - High energy implies probing of short
distances, and production of other, massive particles
24
7 - 14 TeV center of mass energy
VLHCLHC
Gustaaf Brooijmans CERN 2017
Hadron Colliders❖ Incoming longitudinal momentum not known: - “Hard interaction” is between one of the quarks and/or gluons from each
proton, other quarks/gluons are “spectators”
❖ Longitudinal boost “flattens” event to a pancake ➡ We usually work in the plane transverse to the beam
25
Gustaaf Brooijmans CERN 2017
Detectors❖ Make best possible measurement of all particles coming out of collisions
26
Gustaaf Brooijmans CERN 2017
CMS
27
456%3.8T Solenoid
0>?@!76k scintillating PbWO4 crystals A>?@!Scintillator/brass
Interleaved ~7k ch
• Pixels (100x150 µm2) " ~ 1 m2 ~66M ch"• Si Strips (80-180 µm)" ~200 m2 ~9.6M ch!
g8T.#@%o%970>G.7!
5p+_%XQqqfZ!250 Drift Tubes (DT) and 480 Resistive Plate Chambers (RPC)
473 Cathode Strip Chambers (CSC) 432 Resistive Plate Chambers (RPC)
5p+_%f_i4Qg6!
!+&)1%B'#9"&%%%%%%%%%=4<<<%&%CD',)11%2#)('&',%%%=6%(%CD',)11%1'*9&"%%%%%%%;E:5%(%
eq+_%r+sf%
YBO YB1-2
Preshower Si Strips ~16 m2
~137k ch
Foward Cal Steel + quartz Fibers 2~k ch
16
Gustaaf Brooijmans CERN 2017
Charged Particles❖ Combination of pixels, silicon strips (“SCT”) and straw tube transition
radiation tracker (TRT)
28
Gustaaf Brooijmans CERN 2017
Calorimetry❖ Liquid Argon & Pb accordion (EM & forward),
crystals ❖ Scintillator & steel/copper/tungsten (hadronic)
29
Gustaaf Brooijmans CERN 2017
Muons❖ Air-core toroids/flux return; wire chambers
and RPCs
30
Gustaaf Brooijmans CERN 2017
Neutrinos*
31
*(100% acceptance)
Gustaaf Brooijmans CERN 2017 32
✓
✓✓
✓✓✓
✓
✓
✓ ✓ ✓
✓: Detect with high efficiency
✓: Detect by missing transverse energy
✓
✓
✓
✓
✓: Detect through decays: t→Wb, W/Z → leptons, ...
✓
H125500
0
✓
Detecting Particles
Gustaaf Brooijmans CERN 2017
The Work
33
Gustaaf Brooijmans CERN 2017
Steps in a Physics Analysis❖ Choose a topic (often theory-motivated) ❖ What is the final state? ⇒ “Preselection”
- For a search, sufficiently loose to be signal-poor - Prove you understand the detector response, physics processes contributing
- But sufficiently tight to have a manageable data volume - ATLAS/CMS write 1000 Hz × 1+ MB/event = 1+ GB/s
- “4-vectors” is not enough, need some amount of detector info
- In practice, often have preselected sample for frequent analysis, + looser sample for e.g. multijet background with rare passes
❖ Note that data volume ∝ running time, not ∫L34
Gustaaf Brooijmans CERN 2017
Steps (II)❖ Determine preselected sample’s composition - MC and data to understand each contribution
- Multijet background to leptons often extracted from data: rejection factor ~10-4, difficult for simulation to be that accurate
- MC for most other processes, with corrections from data, since generators are (LO,) NLO, NNLO, (LL,) NLL, NNLL
- Also need to correct MC for real-life data conditions - Different alignment, dead channels etc.
- As statistics increase, more difficult, since mis-modelings not hidden by statistical uncertainties anymore
- Mis-modelings often show up in tails
35
Gustaaf Brooijmans CERN 2017
Anecdotes From the Field (I)❖ Everybody wants experimenters to produce results fast - Lots of pressure in the early days of LHC...
36
Only jets, background easy
Gustaaf Brooijmans CERN 2017
Anecdotes From the Field (I)
37
GEANT bug
❖ Everybody wants experimenters to produce results fast - Lots of pressure in the early days of LHC...
Gustaaf Brooijmans CERN 2017
Anecdotes From the Field (I)
38
GEANT bug
Cosmics
❖ Everybody wants experimenters to produce results fast - Lots of pressure in the early days of LHC…
- Sometimes it’s better to take the time needed to understand strange things…
Gustaaf Brooijmans CERN 2017
A Semi-Challenging Search: Higgs to τ µ
39
Gustaaf Brooijmans CERN 2017
❖ 20 fb-1 collected by end 2012 at 8 TeV
40
400000 events in direct productioncan look for rare decays!
Producing Higgses
Gustaaf Brooijmans CERN 2017
Higgs Decay: 125 GeV is Golden
41
Low MassH → bb, ττ, γγ
High MassH → WW, ZZ
Gustaaf Brooijmans CERN 2017
µ+τ❖ Indirect constraints fairly weak (as opposed to e.g. e+µ)
- Indirect: BR(µτ) < ~10%; BR(eµ) < ~10-8
❖ Lepton Flavor remains a mystery - Observing LFV crucial in understanding origin
- Know it exists in the neutrino sector
❖ Experimentally: - With 400k Higgses produced,1% BR yields 4000 signal events (x efficiency)
- Two leptons ⇒ small to moderate background at hadron collider
42
Gustaaf Brooijmans CERN 2017
Tau decays❖ Exploit two channels: - τ→eνν: BR = 18%
- τ→hν: BR = 49% (one charged particle) + 15% (three charged particles)
- Avoid Z → μμ background
❖ Final states are μτe and μτh - Irreducible background is Z → ττ
- Primary discriminating variable is μ-τ invariant mass - Unfortunately not directly reconstructible: neutrinos escape!
43
Gustaaf Brooijmans CERN 2017
Collinear Mass❖ m(H) = 125 GeV, m(τ) = 1.8 GeV ➡ Tau is heavily boosted
➡ Tau decay products are collinear with tau
❖ Under that assumption, know neutrino direction - From direction and missing transverse momentum infer neutrino longitudinal momentum
44
CMS: arXiv:1502.07400PreselectionPhys. Lett. B 749 (2015) 337
Gustaaf Brooijmans CERN 2017
Categorize!❖ Different production mechanism (gluon
fusion vs. vector boson fusion) lead to different topologies
- In practice number of jets
❖ Different decay channels have different reducible backgrounds
- Hadronic tau decays are low multiplicity jets
❖ Categorize to exploit different S/B! - Assign corresponding weights (typically
ln(1+S/B)), to increase sensitivity
45
Even
ts /
10 G
eV
0
1000
2000
3000
4000
5000
6000
7000 eτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherMisidentified leptonsLFV GG Higgs (Br=100%)LFV VBF Higgs (Br=100%)
0-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
10 G
eV
0
1000
2000
3000
4000
5000
6000
7000hτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherτMisidentified
LFV GG Higgs (B=100%)LFV VBF Higgs (B=100%)
(8 TeV)-119.7 fb
CMS 0-Jet
hτµ
[GeV]col
)hτµM(
0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
10 G
eV
0
200
400
600
800
1000
1200
1400
1600
1800
2000eτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherMisidentified leptonsLFV GG Higgs (Br=100%)LFV VBF Higgs (Br=100%)
1-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
10 G
eV
0
500
1000
1500
2000
2500
3000
3500hτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherτMisidentified
LFV GG Higgs (B=100%)LFV VBF Higgs (B=100%)
(8 TeV)-119.7 fb
CMS 1-Jet
hτµ
[GeV]col
)hτµM(
0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
20 G
eV
0
100
200
300
400
500
600
700
800eτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherMisidentified leptonsLFV GG Higgs (Br=100%)LFV VBF Higgs (Br=100%)
2-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
20 G
eV
500
1000
1500
2000
2500
3000hτµData,
Bkgd. uncertaintySM Higgs
ττ→Zt, t, tt
OtherτMisidentified
LFV GG Higgs (B=100%)LFV VBF Higgs (B=100%)
(8 TeV)-119.7 fb
CMS 2-Jet
hτµ
[GeV]col
)hτµM(
0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Gustaaf Brooijmans CERN 2017
Backgrounds❖ Small signal ⇒ need very accurate
background estimate - Use data where possible
❖ In this case: - Z → ττ (irreducible): take Z → µµ events from data,
replace one muon with simulated tau
- Misidentified leptons: get control sample, and independently measure probability to fake e or τh, check in control region
- Rest: simulation46
Even
ts /
10 G
eV
-110
1
10
210
310
410
510 eτµData, Bkgd. stat. uncertaintyBkgd. syst. uncertaintyMisidentified leptonsTrue leptonsLFV GG Higgs (B=100%)LFV VBF Higgs (B=100%)
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Even
ts /
10 G
eV
-110
1
10
210
310
410
510
610hτµData,
Bkgd. stat. uncertainty
Bkgd. syst. uncertainty
τMisidentified
LFV GG Higgs (B=100%)
LFV VBF Higgs (B=100%)
(8 TeV)-119.7 fb
CMS
[GeV]col
)hτµM(
0 100 200 300
Bkgd
Dat
a-Bk
gd
-0.50
0.5
Gustaaf Brooijmans CERN 2017
Finally❖ Tighten cuts and look for signal ❖ Don’t forget systematic uncertainties - Difficult topic: estimators often have known
flaws, but “best we can do”
47
Even
ts /
10 G
eV
0
10
20
30
40
50
60eτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttMisidentified leptonsLFV Higgs (B=0.84%)
0-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.5
Even
ts /
10 G
eV
0
100
200
300
400
500hτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttτMisidentified
LFV Higgs (B=0.84%)
0-Jethτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)hτµM(
0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.5
Even
ts /
10 G
eV
0
10
20
30
40
50
60
70eτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttMisidentified leptonsLFV Higgs (B=0.84%)
1-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.5
Even
ts /
10 G
eV
0
20
40
60
80
100
120
140
160
180hτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttτMisidentified
LFV Higgs (B=0.84%)
1-Jethτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)hτµM(
0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.5
Even
ts /
20 G
eV
0
1
2
3
4
5
6
7
8
9eτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttMisidentified leptonsLFV Higgs (B=0.84%)
2-Jeteτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)eτµM(0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.5
Even
ts /
50 G
eV
0
2
4
6
8
10
12
14
16
18
20
hτµData,
Bkgd. uncertaintySM Higgs
ττ→ZOther
t, t, ttτMisidentified
LFV Higgs (B=0.84%)
2-Jethτµ
(8 TeV)-119.7 fb
CMS
[GeV]col
)hτµM(
0 100 200 300
Bkgd
(fit)
Dat
a-Bk
gd (f
it)
-0.50
0.51
[GeV]col)τµM(
100 200 300
S/(S
+B) W
eigh
ted
Even
ts /
20 G
eV
-4
-2
0
2
4
6
8
10 signal (B=0.84%)τµ→LFV H
Bkgd. uncertainty
Data-Bkgd
(8 TeV)-119.7 fb
CMS
2.5 sigma!
(Correspondsto ~1% BR)
Gustaaf Brooijmans CERN 2017
Run 2
48
CMS-PAS-HIG-17-001
Gustaaf Brooijmans CERN 2017
Higgs Drawbacks❖ So with the addition of a Higgs boson around 125 GeV particle
physics could be “complete” - Like Mendeleev’s table for chemistry, but not understood. By itself, the
Higgs is very unsatisfactory: - Why are the couplings to the fermions what they are?
‣ Dumb luck (aka landscape)?
- What is the link to gravity?
- What about Dark Matter?
- Why does the Higgs break the symmetry?
- Why are there 3....?
49